Release [run release-20261109T112131Z]

#4
by Battam - opened
This view is limited to 50 files because it contains too many changes. See the raw diff here.
Files changed (50) hide show
  1. CHANGELOG.md +50 -0
  2. README.md +32 -29
  3. README.zh.md +32 -29
  4. campaign_files/2026-10-b-disk-images-3/digitalcorpora-nps-edu-deleted/README.md +0 -3
  5. campaign_files/2026-10-g-solver-instances-2/cvrplib/README.md +0 -80
  6. campaign_files/2026-10-g-solver-instances-2/dimacs-graph-coloring-instances/README.md +0 -10
  7. campaign_files/2026-10-g-solver-instances/csplib/README.md +0 -4
  8. campaign_files/2026-10-gt-browser-tests-2/cpython-test-email-data/README.md +0 -6
  9. campaign_files/2026-10-gt-browser-tests-2/crisp-email-reply-parser-fixtures/README.md +0 -6
  10. campaign_files/2026-10-gt-browser-tests-2/data-ical-parse/README.md +0 -4
  11. campaign_files/2026-10-gt-browser-tests-2/data-ical-vcal10/README.md +0 -4
  12. campaign_files/2026-10-gt-browser-tests-2/deltachat-provider-db/README.md +0 -10
  13. campaign_files/2026-10-gt-browser-tests-2/docx2python-test-resources/README.md +0 -6
  14. campaign_files/2026-10-gt-browser-tests-2/dovecot-pigeonhole-tests/README.md +0 -4
  15. campaign_files/2026-10-hw-eda-tests-2/cq-kit-at-commit-2b3897b-github-michaelgale-cq-kit-a-cadquery-helper-0061fa24/README.md +0 -10
  16. campaign_files/2026-10-m-issue-trackers-2/coreboot-coreboot/README.md +0 -4
  17. campaign_files/2026-10-m-issue-trackers-2/coreboot-gerrit-changes/README.md +0 -4
  18. campaign_files/2026-10-m-issue-trackers-2/couchbase-gerrit-changes/README.md +0 -4
  19. campaign_files/2026-10-m-issue-trackers-2/courtney-passenger-gtfs/README.md +0 -4
  20. campaign_files/2026-10-m-issue-trackers-2/dart-gerrit-changes/README.md +0 -4
  21. campaign_files/2026-10-m-issue-trackers-2/dawn-gerrit-changes/README.md +0 -4
  22. campaign_files/2026-10-m-issue-trackers-2/digium-asterisk-users/README.md +0 -4
  23. campaign_files/2026-10-m-issue-trackers-2/django-trac/README.md +0 -4
  24. campaign_files/2026-10-m-issue-trackers-2/dns-oarc-dns-operations/README.md +0 -4
  25. campaign_files/2026-10-m-issue-trackers-2/dotnet-runtime-github-pulls/README.md +0 -4
  26. campaign_files/2026-10-m-issue-trackers-2/dpdk-users/README.md +0 -4
  27. campaign_files/2026-10-m-issue-trackers-2/dragonflybsd-redmine/README.md +0 -6
  28. campaign_files/2026-10-md-media-and-documents/crema-d/README.md +0 -4
  29. campaign_files/2026-10-md-media-and-documents/csl-test-suite-date-datebc/README.md +0 -4
  30. campaign_files/2026-10-md-media-and-documents/csvkit/README.md +0 -4
  31. campaign_files/2026-10-md-media-and-documents/ctex-org-ctex-kit-xecjk-testfiles/README.md +0 -4
  32. campaign_files/2026-10-md-media-and-documents/dcpurton-afesbrand-testfiles/README.md +0 -4
  33. campaign_files/2026-10-md-media-and-documents/dcpurton-ltx-talk-au-testfiles/README.md +0 -4
  34. campaign_files/2026-10-md-media-and-documents/dcpurton-sbltex-bibleref-sbl-testfiles/README.md +0 -4
  35. campaign_files/2026-10-md-media-and-documents/dherse-codly-tests/README.md +0 -4
  36. campaign_files/2026-10-md-media-and-documents/dita-open-toolkit/README.md +0 -4
  37. campaign_files/2026-10-md-media-and-documents/docling-docx-groundtruth/README.md +0 -4
  38. campaign_files/2026-10-md-media-and-documents/docutils-test-functional/README.md +0 -4
  39. campaign_files/2026-10-md-media-and-documents/docutils-test-odt/README.md +0 -4
  40. campaign_files/2026-10-md-media-and-documents/docx-preview/README.md +0 -4
  41. campaign_files/2026-10-md-media-and-documents/doxygen/README.md +0 -4
  42. campaign_files/2026-10-ml-serving-configs-2/cpython-cmath-testcases/README.md +0 -4
  43. campaign_files/2026-10-ml-serving-configs-2/cpython-math-testcases/README.md +0 -4
  44. campaign_files/2026-10-ml-serving-configs-2/crlibm-testdata/README.md +0 -24
  45. campaign_files/2026-10-ml-serving-configs-2/ctranslate2-cpu-benchmark/README.md +0 -4
  46. campaign_files/2026-10-ml-serving-configs-2/ctranslate2-engine-test-expected-values/README.md +0 -8
  47. campaign_files/2026-10-op-rule-calculations-2/dbt-jaffle-shop-seeds/README.md +0 -4
  48. campaign_files/2026-10-op-rule-calculations-2/dib-2019-sandwich-analytics/README.md +0 -4
  49. campaign_files/2026-10-op-rule-calculations-2/dol-whd-flsa-opinion-letters/README.md +0 -6
  50. campaign_files/2026-10-op-rule-calculations-2/dolibarr-phpunit-stock/README.md +0 -4
CHANGELOG.md CHANGED
@@ -2580,3 +2580,53 @@ Changes to the content of this index, oldest first. There is one entry for each
2580
  - concat-stepsize-repro: changed pointer group file data/concat-stepsize-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source concat-stepsize-repro after the commit (row 1 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2581
  - hf-training-records: 18 new name hit(s)
2582
  - npm: 82 new name hit(s)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2580
  - concat-stepsize-repro: changed pointer group file data/concat-stepsize-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source concat-stepsize-repro after the commit (row 1 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2581
  - hf-training-records: 18 new name hit(s)
2582
  - npm: 82 new name hit(s)
2583
+
2584
+ ## 2026-11-09: sources held back
2585
+
2586
+ - Each source below keeps its newest version that passed every check; its newer batch waits until it passes them:
2587
+ - adflib-adf-deleted: changed pointer group file data/adflib-adf-deleted/2026-10-b-disk-images-3/part-00000.parquet: 1 key(s) of the old version are in no data file of source adflib-adf-deleted after the commit (row 2 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2588
+ - adflib-adf-malformed: changed pointer group file data/adflib-adf-malformed/2026-10-b-disk-images-3/part-00000.parquet: 2 key(s) of the old version are in no data file of source adflib-adf-malformed after the commit (row 4 from batch 2026-10-b-disk-images-3, row 5 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2589
+ - akretion-factur-x-libreoffice-extension: changed pointer group file data/akretion-factur-x-libreoffice-extension/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source akretion-factur-x-libreoffice-extension after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2590
+ - austender-ict-procurement-statistics: changed pointer group file data/austender-ict-procurement-statistics/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 2 key(s) of the old version are in no data file of source austender-ict-procurement-statistics after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2, row 2 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2591
+ - australia-government-workbooks: changed pointer group file data/australia-government-workbooks/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 2 key(s) of the old version are in no data file of source australia-government-workbooks after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2, row 2 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2592
+ - australia-government-workbooks-noassertion: changed pointer group file data/australia-government-workbooks-noassertion/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 2 key(s) of the old version are in no data file of source australia-government-workbooks-noassertion after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
2593
+ - basseyisrael-inventory-system: changed pointer group file data/basseyisrael-inventory-system/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source basseyisrael-inventory-system after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2594
+ - bayesian-siac-filter-repro: changed pointer group file data/bayesian-siac-filter-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source bayesian-siac-filter-repro after the commit (row 5 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2595
+ - bbmh-ap-imex-repro: changed pointer group file data/bbmh-ap-imex-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source bbmh-ap-imex-repro after the commit (row 2 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2596
+ - bc-government-workbooks: changed pointer group file data/bc-government-workbooks/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 21 key(s) of the old version are in no data file of source bc-government-workbooks after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers, row 3 from batch 2026-10-a-spreadsheets-with-answers, 18 more); the removal record holds none of them, and none fails a publish check now
2597
+ - bhardwaj2024-excel-gst-billing-invoice: changed pointer group file data/bhardwaj2024-excel-gst-billing-invoice/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source bhardwaj2024-excel-gst-billing-invoice after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2598
+ - bkp-2015-heath-street: changed pointer group file data/bkp-2015-heath-street/2026-10-b-disk-images-3/part-00000.parquet: 1 key(s) of the old version are in no data file of source bkp-2015-heath-street after the commit (row 2 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2599
+ - bonndata-workbooks-cc0: changed pointer group file data/bonndata-workbooks-cc0/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source bonndata-workbooks-cc0 after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2600
+ - borealis-dataverse-workbooks-cc-by: changed pointer group file data/borealis-dataverse-workbooks-cc-by/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 7 key(s) of the old version are in no data file of source borealis-dataverse-workbooks-cc-by after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers, row 3 from batch 2026-10-a-spreadsheets-with-answers, 4 more); the removal record holds none of them, and none fails a publish check now
2601
+ - borealis-dataverse-workbooks-cc0: changed pointer group file data/borealis-dataverse-workbooks-cc0/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 10 key(s) of the old version are in no data file of source borealis-dataverse-workbooks-cc0 after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers, row 3 from batch 2026-10-a-spreadsheets-with-answers, 7 more); the removal record holds none of them, and none fails a publish check now
2602
+ - btrfs-progs-fsck-raw-images: changed pointer group file data/btrfs-progs-fsck-raw-images/2026-10-b-disk-images-3/part-00000.parquet: 6 key(s) of the old version are in no data file of source btrfs-progs-fsck-raw-images after the commit (row 1 from batch 2026-10-b-disk-images-3, row 2 from batch 2026-10-b-disk-images-3, row 3 from batch 2026-10-b-disk-images-3, 3 more); the removal record holds none of them, and none fails a publish check now
2603
+ - btrfs-progs-fuzz-images: changed pointer group file data/btrfs-progs-fuzz-images/2026-10-b-disk-images-3/part-00000.parquet: 15 key(s) of the old version are in no data file of source btrfs-progs-fuzz-images after the commit (row 1 from batch 2026-10-b-disk-images-3, row 2 from batch 2026-10-b-disk-images-3, row 3 from batch 2026-10-b-disk-images-3, 12 more); the removal record holds none of them, and none fails a publish check now
2604
+ - canada-government-workbooks: changed pointer group file data/canada-government-workbooks/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 2 key(s) of the old version are in no data file of source canada-government-workbooks after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers); the removal record holds none of them, and none fails a publish check now
2605
+ - cdi-import-test-images: changed pointer group file data/cdi-import-test-images/2026-10-b-disk-images-3/part-00000.parquet: 3 key(s) of the old version are in no data file of source cdi-import-test-images after the commit (row 3 from batch 2026-10-b-disk-images-3, row 4 from batch 2026-10-b-disk-images-3, row 5 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2606
+ - cfpb-complaints: 1 new name hit(s)
2607
+ - cleancooking-protocols-noassertion: changed pointer group file data/cleancooking-protocols-noassertion/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source cleancooking-protocols-noassertion after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2608
+ - codek11-supply-chain-inventory-analysis: changed pointer group file data/codek11-supply-chain-inventory-analysis/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source codek11-supply-chain-inventory-analysis after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2609
+ - concat-stepsize-repro: changed pointer group file data/concat-stepsize-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source concat-stepsize-repro after the commit (row 1 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2610
+ - convergence-sbp-repro: changed pointer group file data/convergence-sbp-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source convergence-sbp-repro after the commit (row 2 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2611
+ - crosswalk-reproduction-repro: changed pointer group file data/crosswalk-reproduction-repro/2026-10-c-repros/part-00000.parquet: 1 key(s) of the old version are in no data file of source crosswalk-reproduction-repro after the commit (row 2 from batch 2026-10-c-repros); the removal record holds none of them, and none fails a publish check now
2612
+ - data-gov-uk-workbooks: changed pointer group file data/data-gov-uk-workbooks/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 17 key(s) of the old version are in no data file of source data-gov-uk-workbooks after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers, row 3 from batch 2026-10-a-spreadsheets-with-answers, 14 more); the removal record holds none of them, and none fails a publish check now
2613
+ - data-gov-uk-workbooks-noassertion: changed pointer group file data/data-gov-uk-workbooks-noassertion/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 7 key(s) of the old version are in no data file of source data-gov-uk-workbooks-noassertion after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers, row 3 from batch 2026-10-a-spreadsheets-with-answers, 4 more); the removal record holds none of them, and none fails a publish check now
2614
+ - dataversenl-workbooks-cc-by: changed pointer group file data/dataversenl-workbooks-cc-by/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 18 key(s) of the old version are in no data file of source dataversenl-workbooks-cc-by after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers, row 3 from batch 2026-10-a-spreadsheets-with-answers, 15 more); the removal record holds none of them, and none fails a publish check now
2615
+ - dataversenl-workbooks-cc0: changed pointer group file data/dataversenl-workbooks-cc0/2026-10-a-spreadsheets-with-answers/part-00000.parquet: 19 key(s) of the old version are in no data file of source dataversenl-workbooks-cc0 after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers, row 2 from batch 2026-10-a-spreadsheets-with-answers, row 3 from batch 2026-10-a-spreadsheets-with-answers, 16 more); the removal record holds none of them, and none fails a publish check now
2616
+ - dataverseno-workbooks-cc-by: changed pointer group file data/dataverseno-workbooks-cc-by/2026-10-a-spreadsheets-with-answers-2/part-00000.parquet: 2 key(s) of the old version are in no data file of source dataverseno-workbooks-cc-by after the commit (row 1 from batch 2026-10-a-spreadsheets-with-answers-2, row 2 from batch 2026-10-a-spreadsheets-with-answers-2); the removal record holds none of them, and none fails a publish check now
2617
+ - dfir-ntfs-fat-deleted: changed pointer group file data/dfir-ntfs-fat-deleted/2026-10-b-disk-images-2/part-00000.parquet: 2 key(s) of the old version are in no data file of source dfir-ntfs-fat-deleted after the commit (row 3 from batch 2026-10-b-disk-images-2, row 4 from batch 2026-10-b-disk-images-2); the removal record holds none of them, and none fails a publish check now
2618
+ - dftt-carving: changed pointer group file data/dftt-carving/2026-10-b-disk-images-3/part-00000.parquet: 1 key(s) of the old version are in no data file of source dftt-carving after the commit (row 2 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2619
+ - dftt-deleted-files: changed pointer group file data/dftt-deleted-files/2026-10-b-disk-images/part-00000.parquet: 5 key(s) of the old version are in no data file of source dftt-deleted-files after the commit (row 6 from batch 2026-10-b-disk-images, row 7 from batch 2026-10-b-disk-images, row 8 from batch 2026-10-b-disk-images, 2 more); the removal record holds none of them, and none fails a publish check now
2620
+ - dftt-fat-label-anomaly: changed pointer group file data/dftt-fat-label-anomaly/2026-10-b-disk-images-3/part-00000.parquet: 1 key(s) of the old version are in no data file of source dftt-fat-label-anomaly after the commit (row 2 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2621
+ - dftt-partition-tables: changed pointer group file data/dftt-partition-tables/2026-10-b-disk-images-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source dftt-partition-tables after the commit (row 2 from batch 2026-10-b-disk-images-2); the removal record holds none of them, and none fails a publish check now
2622
+ - discutils-iso9660-bad-susp: changed pointer group file data/discutils-iso9660-bad-susp/2026-10-b-disk-images-3/part-00000.parquet: 1 key(s) of the old version are in no data file of source discutils-iso9660-bad-susp after the commit (row 2 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2623
+ - discutils-vhdx-log-replay: changed pointer group file data/discutils-vhdx-log-replay/2026-10-b-disk-images-3/part-00000.parquet: 1 key(s) of the old version are in no data file of source discutils-vhdx-log-replay after the commit (row 2 from batch 2026-10-b-disk-images-3); the removal record holds none of them, and none fails a publish check now
2624
+ - dissect-hypervisor-disk-fixtures: changed pointer group file data/dissect-hypervisor-disk-fixtures/2026-10-b-disk-images-3/part-00000.parquet: 8 key(s) of the old version are in no data file of source dissect-hypervisor-disk-fixtures after the commit (row 9 from batch 2026-10-b-disk-images-3, row 10 from batch 2026-10-b-disk-images-3, row 11 from batch 2026-10-b-disk-images-3, 5 more); the removal record holds none of them, and none fails a publish check now
2625
+ - dod-linear-stability-repro: changed pointer group file data/dod-linear-stability-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source dod-linear-stability-repro after the commit (row 4 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2626
+ - dod-sbp-kinetic-repro: changed pointer group file data/dod-sbp-kinetic-repro/2026-10-c-repros-2/part-00000.parquet: 1 key(s) of the old version are in no data file of source dod-sbp-kinetic-repro after the commit (row 6 from batch 2026-10-c-repros-2); the removal record holds none of them, and none fails a publish check now
2627
+ - dosfstools-fsck-cases: changed pointer group file data/dosfstools-fsck-cases/2026-10-b-disk-images-3/part-00000.parquet: 6 key(s) of the old version are in no data file of source dosfstools-fsck-cases after the commit (row 7 from batch 2026-10-b-disk-images-3, row 8 from batch 2026-10-b-disk-images-3, row 9 from batch 2026-10-b-disk-images-3, 3 more); the removal record holds none of them, and none fails a publish check now
2628
+ - dvn-fvhlgg-repro: changed pointer group file data/dvn-fvhlgg-repro/2026-10-c-repros/part-00000.parquet: 4 key(s) of the old version are in no data file of source dvn-fvhlgg-repro after the commit (row 1 from batch 2026-10-c-repros, row 2 from batch 2026-10-c-repros, row 3 from batch 2026-10-c-repros, 1 more); the removal record holds none of them, and none fails a publish check now
2629
+ - dvn-tkoea1-repro: changed pointer group file data/dvn-tkoea1-repro/2026-10-c-repros/part-00000.parquet: 1 key(s) of the old version are in no data file of source dvn-tkoea1-repro after the commit (row 2 from batch 2026-10-c-repros); the removal record holds none of them, and none fails a publish check now
2630
+ - hf-training-records: 18 new name hit(s)
2631
+ - npm: 82 new name hit(s)
2632
+ - stackexchange-api-posts: 9 new name hit(s)
README.md CHANGED
@@ -24,7 +24,7 @@ configs:
24
  <!-- artifact-pointer-layout: v2 -->
25
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
26
 
27
- ![Public web, in two layers (task material in collections; corpora), one row per file (address, size, SHA-256, license), your selection, fetched and verified](https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.1/docs/banner.svg)
28
 
29
  # Agentic Env Index
30
 
@@ -32,7 +32,7 @@ An index of public files for building agent environments: where each one is, its
32
 
33
  English | [简体中文](README.zh.md)
34
 
35
- [![license: CC0-1.0](https://img.shields.io/badge/license-CC0--1.0-4F6D80)](https://creativecommons.org/publicdomain/zero/1.0/) [![material sets: 1,799](https://img.shields.io/badge/material%20sets-1%2C799-4F6D80)](docs/sources.md) [![corpora: 138](https://img.shields.io/badge/corpora-138-4F6D80)](docs/sources.md) [![release: release-2026-11-09.1](https://img.shields.io/badge/release-release--2026--11--09.1-4F6D80)](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.1)
36
 
37
  ## Dataset Details
38
 
@@ -43,35 +43,38 @@ A material set is a batch of files picked from one source (a site, a repository
43
  - **Task material:** 1,799 material sets in 13 collections, with 4,183 units.
44
  A material set (task-ready) is a batch of files picked from one source (a site, a repository or a dataset), for building tasks;
45
  its files are grouped into units.
46
- - **Corpora:** 138 corpora with 1,235,871,975 rows, to search when no material set fits.
47
  A corpus (plural corpora; raw) is one whole upstream that a collector keeps reading, such as a package registry, a bug tracker, a mailing list or a web archive;
48
  its rows point to files or records, each row one unit.
49
  - **Unit:** A unit is a group of files that its upstream publishes as one thing, delivered whole. Its key, `unit_id`, is the `unit_group` its rows share in `source_meta` (a test case, one workbook, one package), else the row's natural key written as its `uri`, then `@<range_offset>` for a byte range, then `!/<member_path>` for an archive member; it is unique within a source. The README.md and LICENSE the index writes for a collection source belong to the source, not to a unit. `tools/fetch.py` puts every unit in a folder of its own named after its `unit_id` and writes `unit_id` on every line of its manifests. A unit is material, not a task: one unit may give several tasks, and one task may use several units.
50
  - **What it is:** one row per file, with where to download it, its size, its hashes and its license.
51
  The Dataset Viewer has two subsets: `catalog`, one row per material set or corpus with its `kind` and `collection`, and `preview`, the first 10 rows of each.
52
  One material set or corpus is read from the data files its catalog row names in `data_files`, as in the quick start below and with `tools/fetch.py`.
53
- - **Size of the whole index:** Recorded size 4.50 PB; deduplicated size 4.09 PB; size check: 99.7% match (43,648 of the 44,629 rows read compared); not compared: address unreachable 466, file gone 119, login needed 331, no size from the server 61, other error 4; estimated size of 279,909,103 of the 311,970,416 rows without a recorded size: 583 TB (an estimate; 95% range 497 TB to 701 TB, the sum of its sources' ranges, wider than a joint interval; from 15,964 rows read), the other 32,061,313 with too few rows read for an estimate.
54
  The recorded size is the sum of `size_bytes` over the rows that have one (catalog column `bytes`). The check reads a sample of rows of every data file (`meta/sizes.parquet`): 20 of its rows with a recorded size, and of its rows without one its share of its source's sample, sized for their mean size within 25% at 95% confidence by the spread of the sizes read so far (at most 400 rows per source per round, 20 for a source with none read yet); its share is the rows that match their recorded size out of the rows it compared (status `match`, `size_mismatch` or `hash_mismatch`), and the rows it could not compare, which never reached their file, are counted beside it by status (`unreachable`, `gone`, `needs_auth`, `no_length`, `error`; catalog column `size_check`). The estimate for a source's rows without a recorded size (catalog columns `unsized_rows` and `unsized_estimate`) is their number times the mean measured size of such rows read, with a 95% BCa bootstrap range (`scipy.stats.bootstrap`, its default resample count); a source with fewer than 10 such rows read has none. Above one source, the estimate is the sum of its sources' estimates and its range the sum of their ranges, which is wider than a joint interval. The figures of every level are in `meta/size_figures.json`.
55
  - **Who it is for:** people and agents that pick material for terminal tasks and then fetch the files themselves.
56
  - **What it is not:** a list of ready tasks or environments, and not a copy of the files.
57
 
58
  ## Quick start
59
 
60
- Everything below is pinned to the release tag `release-2026-11-09.1`.
61
 
62
  1. Fetch files with one command ([uv](https://docs.astral.sh/uv/) reads the script's dependencies):
63
 
64
  ```sh
65
- uv run https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.1/tools/fetch.py --form spreadsheets --sample 3 --out ./data
66
  ```
67
 
68
  This step returns units: 3 units from each material set of the form `spreadsheets`, each in a folder of its own named after its `unit_id`.
 
 
 
69
 
70
  2. Choose material sets from the catalog with [DuckDB](https://duckdb.org/):
71
 
72
  ```sql
73
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
74
- FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.1/meta/catalog.parquet'
75
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
76
  ORDER BY collection, units DESC;
77
  ```
@@ -82,7 +85,7 @@ with `kind = 'corpus'` it returns corpora instead, counted in rows (`files`).
82
  3. Choose files in one corpus:
83
 
84
  ```sql
85
- SET VARIABLE files = (SELECT list('hf://datasets/Battam/agentic-env-index@release-2026-11-09.1/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.1/meta/catalog.parquet' WHERE source = 'if-archive'));
86
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
87
  FROM read_parquet(getvariable('files'))
88
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
@@ -105,7 +108,7 @@ only their sizes add up to the size of the whole index.
105
  ### Task material, by collection
106
 
107
  1,799 material sets in 13 collections, with 4,183 units.
108
- Recorded size 2.90 GB; deduplicated size 2.82 GB; size check: 99.7% match (8,122 of the 8,144 rows read compared); not compared: address unreachable 10, file gone 2, login needed 8, other error 2.
109
  Upstreams are the upstream works the material sets of a collection come from.
110
 
111
  | Collection | Material sets | Units | Files | Upstreams |
@@ -128,24 +131,24 @@ Upstreams are the upstream works the material sets of a collection come from.
128
 
129
  ### Corpora
130
 
131
- 138 corpora with 1,235,871,975 rows.
132
- Recorded size 4.50 PB; deduplicated size 4.09 PB; size check: 99.7% match (35,526 of the 36,485 rows read compared); not compared: address unreachable 456, file gone 117, login needed 323, no size from the server 61, other error 2; estimated size of 279,909,103 of the 311,970,416 rows without a recorded size: 583 TB (an estimate; 95% range 497 TB to 701 TB, the sum of its sources' ranges, wider than a joint interval; from 15,964 rows read), the other 32,061,313 with too few rows read for an estimate.
133
  Each corpus is counted under its main domain.
134
 
135
  | Domain | Corpora | Rows | Distinct size |
136
  |---|---:|---:|---:|
137
- | Data science | 23 | 750,815,142 | 600 TB |
138
- | System administration | 29 | 185,547,838 | 182 TB |
139
- | Many domains | 16 | 109,251,421 | 224 TB |
140
- | Software engineering | 12 | 65,693,542 | 86.5 TB |
141
- | Science | 15 | 44,541,599 | 251 TB |
142
  | Operations | 6 | 39,404,642 | 1.78 TB |
143
- | Media | 9 | 12,751,384 | 874 TB |
144
- | Machine learning | 6 | 9,520,852 | 1.86 PB |
145
- | Hardware | 5 | 8,715,949 | 1.66 TB |
146
- | Unclassified | 4 | 4,869,099 | 291 GB |
147
- | Security | 10 | 4,519,397 | 11.9 TB |
148
- | Misc | 3 | 241,110 | 2.87 TB |
149
 
150
  - [Data page](https://huggingface.co/spaces/Battam/agentic-env-catalog): browse task material by collection, or corpora, one layer at a time, and build a fetch command.
151
  - [The `catalog` subset in the Dataset Viewer](https://huggingface.co/datasets/Battam/agentic-env-index/viewer/catalog): one row per material set or corpus, with its `kind` and `collection`.
@@ -212,19 +215,19 @@ Every row field and every column of the catalog: [docs/fields.md](docs/fields.md
212
  it contains no personal records of its own.
213
  - Some sources point to material that names its authors, such as mailing lists, bug trackers and forums.
214
  - Rows that point to data about health or about minors carry `health` or `minors` in the row column `sensitive`, so they can be filtered out.
215
- - Personal information read from sample files: 18 of 1,937 sources so far, each from 9 to 30 sample files read on 2026-10-04 and 2026-10-05; by level: `none_known` 3, `present` 14, `suspected` 1. The other 1,919 sources show `not judged yet`. The card of each source gives its level with the sample files it rests on (`risk.pii` and `risk.pii_evidence` in `cards/<source>.yaml`).
216
 
217
  ## Bias, Risks, and Limitations
218
 
219
- - Sizes: 1,890 of 1,937 sources have a size for every row. For the other 47, rows whose size the source does not publish are shown as unknown and left out of size totals, never counted as 0.
220
- - Judged fields: 0 of 1,937 sources have a summary and task fit read from sample files, and 18 of 1,937 have personal information read from sample files (the same count as in the section above); fields not judged yet say so. Counts, types, sizes and licenses of every source come from the index itself.
221
- - Growing sources: 25 of 1,937 sources are still being collected, and their row counts grow with each release.
222
 
223
  ## Updates and Versions
224
 
225
- - Latest release: `release-2026-11-09.1` (its newest CHANGELOG entry is dated 2026-11-09). Releases were published on 14 of the 30 days from 2026-10-11 to 2026-11-09.
226
  - Releases are listed in [CHANGELOG.md](CHANGELOG.md);
227
- each one is a commit with its own release tag (this one is [`release-2026-11-09.1`](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.1)),
228
  so any earlier state stays readable at its tag.
229
  - `meta/projection.json` records the release tag and the version of the program and templates that generated this card.
230
  - Croissant metadata: [generated by the Hub](https://huggingface.co/api/datasets/Battam/agentic-env-index/croissant);
@@ -246,7 +249,7 @@ A removal covers the current version of this repository (the main branch) and th
246
  @misc{agentic_env_index,
247
  title = {agentic-env-index: an index of public materials for building agent environments},
248
  howpublished = {https://huggingface.co/datasets/Battam/agentic-env-index},
249
- note = {Release release-2026-11-09.1},
250
  year = {2026}
251
  }
252
  ```
 
24
  <!-- artifact-pointer-layout: v2 -->
25
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
26
 
27
+ ![Public web, in two layers (task material in collections; corpora), one row per file (address, size, SHA-256, license), your selection, fetched and verified](https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.2/docs/banner.svg)
28
 
29
  # Agentic Env Index
30
 
 
32
 
33
  English | [简体中文](README.zh.md)
34
 
35
+ [![license: CC0-1.0](https://img.shields.io/badge/license-CC0--1.0-4F6D80)](https://creativecommons.org/publicdomain/zero/1.0/) [![material sets: 1,799](https://img.shields.io/badge/material%20sets-1%2C799-4F6D80)](docs/sources.md) [![corpora: 139](https://img.shields.io/badge/corpora-139-4F6D80)](docs/sources.md) [![release: release-2026-11-09.2](https://img.shields.io/badge/release-release--2026--11--09.2-4F6D80)](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.2)
36
 
37
  ## Dataset Details
38
 
 
43
  - **Task material:** 1,799 material sets in 13 collections, with 4,183 units.
44
  A material set (task-ready) is a batch of files picked from one source (a site, a repository or a dataset), for building tasks;
45
  its files are grouped into units.
46
+ - **Corpora:** 139 corpora with 1,216,270,370 rows, to search when no material set fits.
47
  A corpus (plural corpora; raw) is one whole upstream that a collector keeps reading, such as a package registry, a bug tracker, a mailing list or a web archive;
48
  its rows point to files or records, each row one unit.
49
  - **Unit:** A unit is a group of files that its upstream publishes as one thing, delivered whole. Its key, `unit_id`, is the `unit_group` its rows share in `source_meta` (a test case, one workbook, one package), else the row's natural key written as its `uri`, then `@<range_offset>` for a byte range, then `!/<member_path>` for an archive member; it is unique within a source. The README.md and LICENSE the index writes for a collection source belong to the source, not to a unit. `tools/fetch.py` puts every unit in a folder of its own named after its `unit_id` and writes `unit_id` on every line of its manifests. A unit is material, not a task: one unit may give several tasks, and one task may use several units.
50
  - **What it is:** one row per file, with where to download it, its size, its hashes and its license.
51
  The Dataset Viewer has two subsets: `catalog`, one row per material set or corpus with its `kind` and `collection`, and `preview`, the first 10 rows of each.
52
  One material set or corpus is read from the data files its catalog row names in `data_files`, as in the quick start below and with `tools/fetch.py`.
53
+ - **Size of the whole index:** Recorded size 4.45 PB; deduplicated size 4.11 PB; size check: 99.7% match (38,252 of the 39,145 rows read compared); not compared: address unreachable 459, file gone 82, login needed 330, no size from the server 19, other error 3; estimated size of 215,326,513 of the 312,208,580 rows without a recorded size: 202 TB (an estimate; 95% range 171 TB to 245 TB, the sum of its sources' ranges, wider than a joint interval; from 14,161 rows read), the other 96,882,067 with too few rows read for an estimate.
54
  The recorded size is the sum of `size_bytes` over the rows that have one (catalog column `bytes`). The check reads a sample of rows of every data file (`meta/sizes.parquet`): 20 of its rows with a recorded size, and of its rows without one its share of its source's sample, sized for their mean size within 25% at 95% confidence by the spread of the sizes read so far (at most 400 rows per source per round, 20 for a source with none read yet); its share is the rows that match their recorded size out of the rows it compared (status `match`, `size_mismatch` or `hash_mismatch`), and the rows it could not compare, which never reached their file, are counted beside it by status (`unreachable`, `gone`, `needs_auth`, `no_length`, `error`; catalog column `size_check`). The estimate for a source's rows without a recorded size (catalog columns `unsized_rows` and `unsized_estimate`) is their number times the mean measured size of such rows read, with a 95% BCa bootstrap range (`scipy.stats.bootstrap`, its default resample count); a source with fewer than 10 such rows read has none. Above one source, the estimate is the sum of its sources' estimates and its range the sum of their ranges, which is wider than a joint interval. The figures of every level are in `meta/size_figures.json`.
55
  - **Who it is for:** people and agents that pick material for terminal tasks and then fetch the files themselves.
56
  - **What it is not:** a list of ready tasks or environments, and not a copy of the files.
57
 
58
  ## Quick start
59
 
60
+ Everything below is pinned to the release tag `release-2026-11-09.2`.
61
 
62
  1. Fetch files with one command ([uv](https://docs.astral.sh/uv/) reads the script's dependencies):
63
 
64
  ```sh
65
+ uv run https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.2/tools/fetch.py --form spreadsheets --sample 3 --out ./data
66
  ```
67
 
68
  This step returns units: 3 units from each material set of the form `spreadsheets`, each in a folder of its own named after its `unit_id`.
69
+ They are the first 3 of each material set that pass the default profile, in a fixed order: by the sha256 of the path of the data file that holds the unit, then by the unit's key;
70
+ `--limit N` takes the first N units of the whole selection in the same order.
71
+ The script reads the data files in that order and stops once it has the units, so the same command takes the same units of this release.
72
 
73
  2. Choose material sets from the catalog with [DuckDB](https://duckdb.org/):
74
 
75
  ```sql
76
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
77
+ FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.2/meta/catalog.parquet'
78
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
79
  ORDER BY collection, units DESC;
80
  ```
 
85
  3. Choose files in one corpus:
86
 
87
  ```sql
88
+ SET VARIABLE files = (SELECT list('hf://datasets/Battam/agentic-env-index@release-2026-11-09.2/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.2/meta/catalog.parquet' WHERE source = 'if-archive'));
89
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
90
  FROM read_parquet(getvariable('files'))
91
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
 
108
  ### Task material, by collection
109
 
110
  1,799 material sets in 13 collections, with 4,183 units.
111
+ Recorded size 2.90 GB; deduplicated size 2.82 GB; size check: 99.8% match (7,529 of the 7,551 rows read compared); not compared: address unreachable 10, file gone 2, login needed 8, other error 2.
112
  Upstreams are the upstream works the material sets of a collection come from.
113
 
114
  | Collection | Material sets | Units | Files | Upstreams |
 
131
 
132
  ### Corpora
133
 
134
+ 139 corpora with 1,216,270,370 rows.
135
+ Recorded size 4.45 PB; deduplicated size 4.11 PB; size check: 99.6% match (30,723 of the 31,594 rows read compared); not compared: address unreachable 449, file gone 80, login needed 322, no size from the server 19, other error 1; estimated size of 215,326,513 of the 312,208,580 rows without a recorded size: 202 TB (an estimate; 95% range 171 TB to 245 TB, the sum of its sources' ranges, wider than a joint interval; from 14,161 rows read), the other 96,882,067 with too few rows read for an estimate.
136
  Each corpus is counted under its main domain.
137
 
138
  | Domain | Corpora | Rows | Distinct size |
139
  |---|---:|---:|---:|
140
+ | Data science | 23 | 750,815,212 | 600 TB |
141
+ | System administration | 29 | 185,826,217 | 187 TB |
142
+ | Many domains | 16 | 109,262,742 | 224 TB |
143
+ | Software engineering | 12 | 45,267,701 | 87.3 TB |
144
+ | Science | 15 | 44,541,917 | 251 TB |
145
  | Operations | 6 | 39,404,642 | 1.78 TB |
146
+ | Media | 9 | 12,751,612 | 874 TB |
147
+ | Machine learning | 7 | 9,636,889 | 1.86 PB |
148
+ | Hardware | 5 | 8,716,520 | 1.67 TB |
149
+ | Unclassified | 4 | 5,284,162 | 997 GB |
150
+ | Security | 10 | 4,521,631 | 11.9 TB |
151
+ | Misc | 3 | 241,125 | 2.87 TB |
152
 
153
  - [Data page](https://huggingface.co/spaces/Battam/agentic-env-catalog): browse task material by collection, or corpora, one layer at a time, and build a fetch command.
154
  - [The `catalog` subset in the Dataset Viewer](https://huggingface.co/datasets/Battam/agentic-env-index/viewer/catalog): one row per material set or corpus, with its `kind` and `collection`.
 
215
  it contains no personal records of its own.
216
  - Some sources point to material that names its authors, such as mailing lists, bug trackers and forums.
217
  - Rows that point to data about health or about minors carry `health` or `minors` in the row column `sensitive`, so they can be filtered out.
218
+ - Personal information read from sample files: 18 of 1,938 sources so far, each from 9 to 30 sample files read on 2026-10-04 and 2026-10-05; by level: `none_known` 3, `present` 14, `suspected` 1. The other 1,920 sources show `not judged yet`. The card of each source gives its level with the sample files it rests on (`risk.pii` and `risk.pii_evidence` in `cards/<source>.yaml`).
219
 
220
  ## Bias, Risks, and Limitations
221
 
222
+ - Sizes: 1,891 of 1,938 sources have a size for every row. For the other 47, rows whose size the source does not publish are shown as unknown and left out of size totals, never counted as 0.
223
+ - Judged fields: 0 of 1,938 sources have a summary and task fit read from sample files, and 18 of 1,938 have personal information read from sample files (the same count as in the section above); fields not judged yet say so. Counts, types, sizes and licenses of every source come from the index itself.
224
+ - Growing sources: 25 of 1,938 sources are still being collected, and their row counts grow with each release.
225
 
226
  ## Updates and Versions
227
 
228
+ - Latest release: `release-2026-11-09.2` (its newest CHANGELOG entry is dated 2026-11-09). Releases were published on 14 of the 30 days from 2026-10-11 to 2026-11-09.
229
  - Releases are listed in [CHANGELOG.md](CHANGELOG.md);
230
+ each one is a commit with its own release tag (this one is [`release-2026-11-09.2`](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.2)),
231
  so any earlier state stays readable at its tag.
232
  - `meta/projection.json` records the release tag and the version of the program and templates that generated this card.
233
  - Croissant metadata: [generated by the Hub](https://huggingface.co/api/datasets/Battam/agentic-env-index/croissant);
 
249
  @misc{agentic_env_index,
250
  title = {agentic-env-index: an index of public materials for building agent environments},
251
  howpublished = {https://huggingface.co/datasets/Battam/agentic-env-index},
252
+ note = {Release release-2026-11-09.2},
253
  year = {2026}
254
  }
255
  ```
README.zh.md CHANGED
@@ -11,7 +11,7 @@ tags:
11
  ---
12
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
13
 
14
- ![公开网络,分两层(按合集的任务材料;语料库),每个文件一行(地址、大小、SHA-256、许可证),你的选择,下载并校验](https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.1/docs/banner.zh.svg)
15
 
16
  # Agentic Env Index
17
 
@@ -19,7 +19,7 @@ tags:
19
 
20
  [English](README.md) | 简体中文
21
 
22
- [![许可证: CC0-1.0](https://img.shields.io/badge/%E8%AE%B8%E5%8F%AF%E8%AF%81-CC0--1.0-4F6D80)](https://creativecommons.org/publicdomain/zero/1.0/) [![材料集: 1,799](https://img.shields.io/badge/%E6%9D%90%E6%96%99%E9%9B%86-1%2C799-4F6D80)](docs/sources.zh.md) [![语料库: 138](https://img.shields.io/badge/%E8%AF%AD%E6%96%99%E5%BA%93-138-4F6D80)](docs/sources.zh.md) [![版本: release-2026-11-09.1](https://img.shields.io/badge/%E7%89%88%E6%9C%AC-release--2026--11--09.1-4F6D80)](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.1)
23
 
24
  ## 数据集详情
25
 
@@ -30,35 +30,38 @@ tags:
30
  - **任务材料:** 13 个合集中的 1,799 个材料集,共 4,183 个单元。
31
  材料集(任务就绪)是从同一个来源挑出来的一批文件(一个网站、一个仓库或一个数据集),用来构建任务;
32
  它的文件分组为单元。
33
- - **语料库:** 138 个语料库,共 1,235,871,975 行,在没有合适的材料集时用来搜索。
34
  语料库(原始)是采集器持续读取的一个完整上游,例如软件包仓库、缺陷跟踪系统、邮件列表或网页存档;
35
  它的行指向文件或记录,每行是一个单元。
36
  - **单元:** 单元是上游作为一件东西发布的一组文件,整体交付。它的键 `unit_id` 是它的各行在 `source_meta` 中共用的 `unit_group`(一个测试用例、一个工作簿、一个软件包),否则是该行的自然键,写作它的 `uri`,字节范围再接 `@<range_offset>`,存档成员再接 `!/<member_path>`;它在一个来源内唯一。索引为集合来源写的 README.md 和 LICENSE 属于来源,不属于任何单元。`tools/fetch.py` 把每个单元放进以它的 `unit_id` 命名的单独文件夹,并在它的清单的每一行写上 `unit_id`。单元是材料,不是题:一个单元可以出几道题,一道题也可以用几个单元。
37
  - **是什么:** 每个文件一行,写明下载地址、大小、哈希值和许可证。
38
  数据集查看器有两个子集:`catalog`,每个材料集或语料库一行,带有它的 `kind` 和 `collection`;`preview`,每个的前 10 行。
39
  单个材料集或语料库从它在目录中那一行的 `data_files` 列出的数据文件读取,见下面的快速上手和 `tools/fetch.py`。
40
- - **整个索引的大小:** 记录的大小 4.50 PB;去重后大小 4.09 PB;大小核对:99.7% 一致(读过 44,629 行,比对了 43,648 行);未比对:地址无法访问 466 行、文件已不存在 119 行、需要登录 331 行、服务器未给出大小 61 行、其他错误 4 行;没有记录大小的 311,970,416 行中 279,909,103 行的估计大小:583 TB(这是估计值;95% 范围 497 TB 到 701 TB,为其各来源范围之和,比联合区间更宽;依据读过的 15,964 行),其余 32,061,313 行读过的太少,无法估计。
41
  记录的大小是有大小的各行 `size_bytes` 之和(目录列 `bytes`)。核对读取���个数据文件中抽取的一组行(`meta/sizes.parquet`):有记录大小的行抽 20 行;没有记录大小的行按其所在来源的样本分摊,样本量按目前读到的大小的离散程度定,使其平均大小在 95% 置信度下误差不超过 25%(每个来源每轮最多 400 行,还没有读数的来源取 20 行);其比例是比对过的行(状态 `match`、`size_mismatch` 或 `hash_mismatch`)中与记录的大小一致的行所占的比例,没能比对的行(没有取到文件)按状态另行计数(`unreachable`、`gone`、`needs_auth`、`no_length`、`error`;目录列 `size_check`)。一个来源没有记录大小的行的估计(目录列 `unsized_rows` 和 `unsized_estimate`)是这些行的行数乘以读过的这类行的平均实测大小,并给出 95% BCa 自助法范围(`scipy.stats.bootstrap`,用其默认重抽样次数);读过的这类行少于 10 行的来源没有估计。在来源之上的层级,估计是其各来源估计之和,范围是各来源范围之和,比联合区间更宽。每个层级的数字都在 `meta/size_figures.json` 中。
42
  - **给谁用:** 为终端任务挑选材料、然后自己下载文件的人和智能体。
43
  - **不是什么:** 不是现成的任务或环境列表,也不是文件的副本。
44
 
45
  ## 快速上手
46
 
47
- 下面的一切都固定在发布标签 `release-2026-11-09.1` 上。
48
 
49
  1. 用一条命令下载文件([uv](https://docs.astral.sh/uv/) 会读取脚本声明的依赖):
50
 
51
  ```sh
52
- uv run https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.1/tools/fetch.py --form spreadsheets --sample 3 --out ./data
53
  ```
54
 
55
  这一步返回单元:形式为 `spreadsheets` 的每个材料集各 3 个单元,每个单元放在以它的 `unit_id` 命名的单独文件夹里。
 
 
 
56
 
57
  2. 用 [DuckDB](https://duckdb.org/) 从目录中挑选材料集:
58
 
59
  ```sql
60
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
61
- FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.1/meta/catalog.parquet'
62
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
63
  ORDER BY collection, units DESC;
64
  ```
@@ -69,7 +72,7 @@ ORDER BY collection, units DESC;
69
  3. 在一个语料库里挑选文件:
70
 
71
  ```sql
72
- SET VARIABLE files = (SELECT list('hf://datasets/Battam/agentic-env-index@release-2026-11-09.1/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.1/meta/catalog.parquet' WHERE source = 'if-archive'));
73
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
74
  FROM read_parquet(getvariable('files'))
75
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
@@ -92,7 +95,7 @@ ORDER BY size_bytes LIMIT 20;
92
  ### 任务材料,按合集
93
 
94
  13 个合集中的 1,799 个材料集,共 4,183 个单元。
95
- 记录的大小 2.90 GB;去重后大小 2.82 GB;大小核对:99.7% 一致(读过 8,144 行,比对了 8,122 行);未比对:地址无法访问 10 行、文件已不存在 2 行、需要登录 8 行、其他错误 2 行。
96
  上游是一个合集的材料集所来自的上游作品。
97
 
98
  | 合集 | 材料集数 | 单元数 | 文件数 | 上游数 |
@@ -115,24 +118,24 @@ ORDER BY size_bytes LIMIT 20;
115
 
116
  ### 语料库
117
 
118
- 138 个语料库,共 1,235,871,975 行。
119
- 记录的大小 4.50 PB;去重后大小 4.09 PB;大小核对:99.7% 一致(读过 36,485 行,比对了 35,526 行);未比对:地址无法访问 456 行、文件已不存在 117 行、需要登录 323 行、服务器未给出大小 61 行、其他错误 2 行;没有记录大小的 311,970,416 行中 279,909,103 行的估计大小:583 TB(这是估计值;95% 范围 497 TB 到 701 TB,为其各来源范围之和,比联合区间更宽;依据读过的 15,964 行),其��� 32,061,313 行读过的太少,无法估计。
120
  每个语料库计入它的主要领域。
121
 
122
  | 领域 | 语料库数 | 行数 | 去重后大小 |
123
  |---|---:|---:|---:|
124
- | 数据科学 / 数据处理 / 查询 | 23 | 750,815,142 | 600 TB |
125
- | 系统管理 / DevOps / 环境构建 | 29 | 185,547,838 | 182 TB |
126
- | 多领域 | 16 | 109,251,421 | 224 TB |
127
- | 软件工程 | 12 | 65,693,542 | 86.5 TB |
128
- | 科学 | 15 | 44,541,599 | 251 TB |
129
  | 业务运营流程 | 6 | 39,404,642 | 1.78 TB |
130
- | 媒体 | 9 | 12,751,384 | 874 TB |
131
- | 机器学习 / 模型运维 | 6 | 9,520,852 | 1.86 PB |
132
- | 硬件 | 5 | 8,715,949 | 1.66 TB |
133
- | 未分类 | 4 | 4,869,099 | 291 GB |
134
- | 安全 | 10 | 4,519,397 | 11.9 TB |
135
- | 杂项 / 游戏 / 策略 / 约束调度 | 3 | 241,110 | 2.87 TB |
136
 
137
  - [数据页面](https://huggingface.co/spaces/Battam/agentic-env-catalog):按合集浏览任务材料,或浏览语料库,每次一层,并生成下载命令。
138
  - [数据集查看器中的 `catalog` 子集](https://huggingface.co/datasets/Battam/agentic-env-index/viewer/catalog):每个材料集或语料库一行,带有它的 `kind` 和 `collection`。
@@ -200,19 +203,19 @@ ORDER BY size_bytes LIMIT 20;
200
  它本身不含任何个人记录。
201
  - 有些来源指向署有作者姓名的材料,例如邮件列表、缺陷跟踪系统和论坛。
202
  - 指向健康或未成年人相关数据的行,在行列 `sensitive` 中标有 `health` 或 `minors`,可以据此过滤掉。
203
- - 从样本文件读出的个人信息:目前 1,937 个来源中已判读 18 个,每个来源读了 9 到 30 个样本文件,读取日期为 2026-10-04 和 2026-10-05;按等级:`none_known` 3, `present` 14, `suspected` 1。其余 1,919 个来源显示 `not judged yet`(尚未判读)。每个来源的卡片给出它的等级和所依据的样本文件(`cards/<source>.yaml` 里的 `risk.pii` 和 `risk.pii_evidence`)。
204
 
205
  ## 偏差、风险与局限
206
 
207
- - 大小:1,937 个来源中有 1,890 个每一行都有大小。其余 47 个来源里,来源没有公布大小的行显示为未知,不计入大小合计,也从不按 0 计。
208
- - 判读字段:1,937 个来源中有 0 个已从样本文件读出摘要和任务适配度,1,937 个中有 18 个已从样本文件读出个人信息(与上一节的计数相同);尚未判读的字段会注明。每个来源的计数、类型、大小和许可证都直接来自索引本身。
209
- - 仍在增长的来源:1,937 个来源中有 25 个仍在收集中,它们的行数随每次发布增加。
210
 
211
  ## 更新与版本
212
 
213
- - 最新发布:`release-2026-11-09.1`(其最新的 CHANGELOG 条目日期为 2026-11-09)。 从 2026-10-11 到 2026-11-09 的 30 天里,有 14 天发布过新版本。
214
  - 各���发布列在 [CHANGELOG.md](CHANGELOG.md) 中;
215
- 每次发布是一个带有自己发布标签的提交(本次为 [`release-2026-11-09.1`](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.1)),
216
  因此任何较早的状态都可以在它的标签处读到。
217
  - `meta/projection.json` 记录发布标签,以及生成本卡片的程序和模板的版本。
218
  - Croissant 元数据:[由 Hub 生成](https://huggingface.co/api/datasets/Battam/agentic-env-index/croissant);
@@ -242,7 +245,7 @@ ORDER BY size_bytes LIMIT 20;
242
  @misc{agentic_env_index,
243
  title = {agentic-env-index: an index of public materials for building agent environments},
244
  howpublished = {https://huggingface.co/datasets/Battam/agentic-env-index},
245
- note = {Release release-2026-11-09.1},
246
  year = {2026}
247
  }
248
  ```
 
11
  ---
12
  <!-- Generated by artpool.index_projection from meta/, cards/*.yaml, the field registry and its templates. Every release regenerates it, so edits to this file do not last. -->
13
 
14
+ ![公开网络,分两层(按合集的任务材料;语料库),每个文件一行(地址、大小、SHA-256、许可证),你的选择,下载并校验](https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.2/docs/banner.zh.svg)
15
 
16
  # Agentic Env Index
17
 
 
19
 
20
  [English](README.md) | 简体中文
21
 
22
+ [![许可证: CC0-1.0](https://img.shields.io/badge/%E8%AE%B8%E5%8F%AF%E8%AF%81-CC0--1.0-4F6D80)](https://creativecommons.org/publicdomain/zero/1.0/) [![材料集: 1,799](https://img.shields.io/badge/%E6%9D%90%E6%96%99%E9%9B%86-1%2C799-4F6D80)](docs/sources.zh.md) [![语料库: 139](https://img.shields.io/badge/%E8%AF%AD%E6%96%99%E5%BA%93-139-4F6D80)](docs/sources.zh.md) [![版本: release-2026-11-09.2](https://img.shields.io/badge/%E7%89%88%E6%9C%AC-release--2026--11--09.2-4F6D80)](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.2)
23
 
24
  ## 数据集详情
25
 
 
30
  - **任务材料:** 13 个合集中的 1,799 个材料集,共 4,183 个单元。
31
  材料集(任务就绪)是从同一个来源挑出来的一批文件(一个网站、一个仓库或一个数据集),用来构建任务;
32
  它的文件分组为单元。
33
+ - **语料库:** 139 个语料库,共 1,216,270,370 行,在没有合适的材料集时用来搜索。
34
  语料库(原始)是采集器持续读取的一个完整上游,例如软件包仓库、缺陷跟踪系统、邮件列表或网页存档;
35
  它的行指向文件或记录,每行是一个单元。
36
  - **单元:** 单元是上游作为一件东西发布的一组文件,整体交付。它的键 `unit_id` 是它的各行在 `source_meta` 中共用的 `unit_group`(一个测试用例、一个工作簿、一个软件包),否则是该行的自然键,写作它的 `uri`,字节范围再接 `@<range_offset>`,存档成员再接 `!/<member_path>`;它在一个来源内唯一。索引为集合来源写的 README.md 和 LICENSE 属于来源,不属于任何单元。`tools/fetch.py` 把每个单元放进以它的 `unit_id` 命名的单独文件夹,并在它的清单的每一行写上 `unit_id`。单元是材料,不是题:一个单元可以出几道题,一道题也可以用几个单元。
37
  - **是什么:** 每个文件一行,写明下载地址、大小、哈希值和许可证。
38
  数据集查看器有两个子集:`catalog`,每个材料集或语料库一行,带有它的 `kind` 和 `collection`;`preview`,每个的前 10 行。
39
  单个材料集或语料库从它在目录中那一行的 `data_files` 列出的数据文件读取,见下面的快速上手和 `tools/fetch.py`。
40
+ - **整个索引的大小:** 记录的大小 4.45 PB;去重后大小 4.11 PB;大小核对:99.7% 一致(读过 39,145 行,比对了 38,252 行);未比对:地址无法访问 459 行、文件已不存在 82 行、需要登录 330 行、服务器未给出大小 19 行、其他错误 3 行;没有记录大小的 312,208,580 行中 215,326,513 行的估计大小:202 TB(这是估计值;95% 范围 171 TB 到 245 TB,为其各来源范围之和,比联合区间更宽;依据读过的 14,161 行),其余 96,882,067 行读过的太少,无法估计。
41
  记录的大小是有大小的各行 `size_bytes` 之和(目录列 `bytes`)。核对读取���个数据文件中抽取的一组行(`meta/sizes.parquet`):有记录大小的行抽 20 行;没有记录大小的行按其所在来源的样本分摊,样本量按目前读到的大小的离散程度定,使其平均大小在 95% 置信度下误差不超过 25%(每个来源每轮最多 400 行,还没有读数的来源取 20 行);其比例是比对过的行(状态 `match`、`size_mismatch` 或 `hash_mismatch`)中与记录的大小一致的行所占的比例,没能比对的行(没有取到文件)按状态另行计数(`unreachable`、`gone`、`needs_auth`、`no_length`、`error`;目录列 `size_check`)。一个来源没有记录大小的行的估计(目录列 `unsized_rows` 和 `unsized_estimate`)是这些行的行数乘以读过的这类行的平均实测大小,并给出 95% BCa 自助法范围(`scipy.stats.bootstrap`,用其默认重抽样次数);读过的这类行少于 10 行的来源没有估计。在来源之上的层级,估计是其各来源估计之和,范围是各来源范围之和,比联合区间更宽。每个层级的数字都在 `meta/size_figures.json` 中。
42
  - **给谁用:** 为终端任务挑选材料、然后自己下载文件的人和智能体。
43
  - **不是什么:** 不是现成的任务或环境列表,也不是文件的副本。
44
 
45
  ## 快速上手
46
 
47
+ 下面的一切都固定在发布标签 `release-2026-11-09.2` 上。
48
 
49
  1. 用一条命令下载文件([uv](https://docs.astral.sh/uv/) 会读取脚本声明的依赖):
50
 
51
  ```sh
52
+ uv run https://huggingface.co/datasets/Battam/agentic-env-index/resolve/release-2026-11-09.2/tools/fetch.py --form spreadsheets --sample 3 --out ./data
53
  ```
54
 
55
  这一步返回单元:形式为 `spreadsheets` 的每个材料集各 3 个单元,每个单元放在以它的 `unit_id` 命名的单独文件夹里。
56
+ 它们是每个材料集中通过默认配置、按固定顺序排在最前的 3 个单元:先按存放该单元的数据文件路径的 sha256 排,再按单元的键排;
57
+ `--limit N` 按同一顺序取整个选择中的前 N 个单元。
58
+ 脚本按这个顺序读取数据文件,取够单元就停下,所以同一条命令在这个发布上总是取到同样的单元。
59
 
60
  2. 用 [DuckDB](https://duckdb.org/) 从目录中挑选材料集:
61
 
62
  ```sql
63
  SELECT source, collection, title, units, files, top_types, license.tier AS license_tier, needs_auth
64
+ FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.2/meta/catalog.parquet'
65
  WHERE kind = 'material_set' AND needs_auth = 'no' AND license.tier IN ('permissive', 'share-alike')
66
  ORDER BY collection, units DESC;
67
  ```
 
72
  3. 在一个语料库里挑选文件:
73
 
74
  ```sql
75
+ SET VARIABLE files = (SELECT list('hf://datasets/Battam/agentic-env-index@release-2026-11-09.2/' || f) FROM (SELECT unnest(data_files) AS f FROM 'hf://datasets/Battam/agentic-env-index@release-2026-11-09.2/meta/catalog.parquet' WHERE source = 'if-archive'));
76
  SELECT uri, size_bytes, lower(hex(sha256)) AS sha256
77
  FROM read_parquet(getvariable('files'))
78
  WHERE source = 'if-archive' AND ext = 'pdf' AND size_bytes BETWEEN 100000 AND 5000000
 
95
  ### 任务材料,按合集
96
 
97
  13 个合集中的 1,799 个材料集,共 4,183 个单元。
98
+ 记录的大小 2.90 GB;去重后大小 2.82 GB;大小核对:99.8% 一致(读过 7,551 行,比对了 7,529 行);未比对:地址无法访问 10 行、文件已不存在 2 行、需要登录 8 行、其他错误 2 行。
99
  上游是一个合集的材料集所来自的上游作品。
100
 
101
  | 合集 | 材料集数 | 单元数 | 文件数 | 上游数 |
 
118
 
119
  ### 语料库
120
 
121
+ 139 个语料库,共 1,216,270,370 行。
122
+ 记录的大小 4.45 PB;去重后大小 4.11 PB;大小核对:99.6% 一致(读过 31,594 行,比对了 30,723 行);未比对:地址无法访问 449 行、文件已不存在 80 行、需要登录 322 行、服务器未给出大小 19 行、其他错误 1 行;没有记录大小的 312,208,580 行中 215,326,513 行的估计大小:202 TB(这是估计值;95% 范围 171 TB 到 245 TB,为其各来源范围之和,比联合区间更宽;依据读过的 14,161 行),其余 96,882,067 行读过的太少,无法估计。
123
  每个语料库计入它的主要领域。
124
 
125
  | 领域 | 语料库数 | 行数 | 去重后大小 |
126
  |---|---:|---:|---:|
127
+ | 数据科学 / 数据处理 / 查询 | 23 | 750,815,212 | 600 TB |
128
+ | 系统管理 / DevOps / 环境构建 | 29 | 185,826,217 | 187 TB |
129
+ | 多领域 | 16 | 109,262,742 | 224 TB |
130
+ | 软件工程 | 12 | 45,267,701 | 87.3 TB |
131
+ | 科学 | 15 | 44,541,917 | 251 TB |
132
  | 业务运营流程 | 6 | 39,404,642 | 1.78 TB |
133
+ | 媒体 | 9 | 12,751,612 | 874 TB |
134
+ | 机器学习 / 模型运维 | 7 | 9,636,889 | 1.86 PB |
135
+ | 硬件 | 5 | 8,716,520 | 1.67 TB |
136
+ | 未分类 | 4 | 5,284,162 | 997 GB |
137
+ | 安全 | 10 | 4,521,631 | 11.9 TB |
138
+ | 杂项 / 游戏 / 策略 / 约束调度 | 3 | 241,125 | 2.87 TB |
139
 
140
  - [数据页面](https://huggingface.co/spaces/Battam/agentic-env-catalog):按合集浏览任务材料,或浏览语料库,每次一层,并生成下载命令。
141
  - [数据集查看器中的 `catalog` 子集](https://huggingface.co/datasets/Battam/agentic-env-index/viewer/catalog):每个材料集或语料库一行,带有它的 `kind` 和 `collection`。
 
203
  它本身不含任何个人记录。
204
  - 有些来源指向署有作者姓名的材料,例如邮件列表、缺陷跟踪系统和论坛。
205
  - 指向健康或未成年人相关数据的行,在行列 `sensitive` 中标有 `health` 或 `minors`,可以据此过滤掉。
206
+ - 从样本文件读出的个人信息:目前 1,938 个来源中已判读 18 个,每个来源读了 9 到 30 个样本文件,读取日期为 2026-10-04 和 2026-10-05;按等级:`none_known` 3, `present` 14, `suspected` 1。其余 1,920 个来源显示 `not judged yet`(尚未判读)。每个来源的卡片给出它的等级和所依据的样本文件(`cards/<source>.yaml` 里的 `risk.pii` 和 `risk.pii_evidence`)。
207
 
208
  ## 偏差、风险与局限
209
 
210
+ - 大小:1,938 个来源中有 1,891 个每一行都有大小。其余 47 个来源里,来源没有公布大小的行显示为未知,不计入大小合计,也从不按 0 计。
211
+ - 判读字段:1,938 个来源中有 0 个已从样本文件读出摘要和任务适配度,1,938 个中有 18 个已从样本文件读出个人信息(与上一节的计数相同);尚未判读的字段会注明。每个来源的计数、类型、大小和许可证都直接来自索引本身。
212
+ - 仍在增长的来源:1,938 个来源中有 25 个仍在收集中,它们的行数随每次发布增加。
213
 
214
  ## 更新与版本
215
 
216
+ - 最新发布:`release-2026-11-09.2`(其最新的 CHANGELOG 条目日期为 2026-11-09)。 从 2026-10-11 到 2026-11-09 的 30 天里,有 14 天发布过新版本。
217
  - 各���发布列在 [CHANGELOG.md](CHANGELOG.md) 中;
218
+ 每次发布是一个带有自己发布标签的提交(本次为 [`release-2026-11-09.2`](https://huggingface.co/datasets/Battam/agentic-env-index/tree/release-2026-11-09.2)),
219
  因此任何较早的状态都可以在它的标签处读到。
220
  - `meta/projection.json` 记录发布标签,以及生成本卡片的程序和模板的版本。
221
  - Croissant 元数据:[由 Hub 生成](https://huggingface.co/api/datasets/Battam/agentic-env-index/croissant);
 
245
  @misc{agentic_env_index,
246
  title = {agentic-env-index: an index of public materials for building agent environments},
247
  howpublished = {https://huggingface.co/datasets/Battam/agentic-env-index},
248
+ note = {Release release-2026-11-09.2},
249
  year = {2026}
250
  }
251
  ```
campaign_files/2026-10-b-disk-images-3/digitalcorpora-nps-edu-deleted/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: deleted_files.
12
-
13
  **Tools to open and check:** Open with tools for these file types: .e01.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -19,5 +17,4 @@
19
  Files:
20
 
21
  - `images/gilbert1.E01`: Nps-2009-edu-corrupt1: gilbert1.E01 (student USB stick, FAT16, deleted readMe files and video copies).
22
- Reference result: deleted_files
23
  Dependencies: sleuthkit with libewf
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Open with tools for these file types: .e01.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
17
  Files:
18
 
19
  - `images/gilbert1.E01`: Nps-2009-edu-corrupt1: gilbert1.E01 (student USB stick, FAT16, deleted readMe files and video copies).
 
20
  Dependencies: sleuthkit with libewf
campaign_files/2026-10-g-solver-instances-2/cvrplib/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 39 of 39 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Best known total cost plus reference routes (proven optimal); best known total cost plus reference routes (upper bound, not proven optimal); best known total distance plus reference routes (proven optimal); best known total distance plus reference routes (upper bound); optimal total distance plus reference routes; best known total distance plus reference routes (not proven optimal); best known (not proven optimal) total distance plus reference routes. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Tools are named in the check written per file in the notes.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,312 +19,234 @@ Files:
21
  - `cvrplib/A-n32-k5.vrp`: CVRP instance A-n32-k5 (31 customers) with best known solution 784.
22
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
23
  Origin: published benchmark instance from Augerat et al. (1995) set A, as served by CVRPLIB; recorded tool_fixture
24
- Reference result: best known total cost plus reference routes (proven optimal)
25
- Reference source: CVRPLIB BKS file A-n32-k5.sol (COMMENT line 2 of A-n32-k5.vrp says "Optimal value: 784"; the CVRPLIB index marks it optimal).
26
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in A-n32-k5.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 784, the published best known solution (proven optimal, so a lower total means a checker error)
27
  Companions (not included): cvrplib/A-n32-k5.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
28
  Dependencies: no upstream code used
29
  - `cvrplib/X-n101-k25.vrp`: CVRP instance X-n101-k25 (100 customers) with best known solution 27591.
30
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
31
  Origin: published benchmark instance from Uchoa et al. (2014) set X, as served by CVRPLIB; recorded tool_fixture
32
- Reference result: best known total cost plus reference routes (proven optimal)
33
- Reference source: CVRPLIB BKS file X-n101-k25.sol (X-n101-k25 is listed with Opt in the CVRPLIB index; .sol line 27 "Cost 27591").
34
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in X-n101-k25.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 27591, the published best known solution (proven optimal, so a lower total means a checker error)
35
  Companions (not included): cvrplib/X-n101-k25.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
36
  Dependencies: no upstream code used
37
  - `cvrplib/CMT1.vrp`: CVRP instance CMT1 (50 customers) with best known solution 524.611.
38
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
39
  Origin: published benchmark instance from Christofides, Mingozzi and Toth (1979), as served by CVRPLIB; recorded tool_fixture
40
- Reference result: best known total cost plus reference routes (proven optimal)
41
- Reference source: CVRPLIB BKS file CMT1.sol (.sol line 6 "Cost 524.611" (COMMENT line 2 of CMT1.vrp: 524.61)).
42
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in CMT1.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 524.611 plus 0.0005, the published best known solution (proven optimal, so a lower total means a checker error)
43
  Companions (not included): cvrplib/CMT1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
44
  Dependencies: no upstream code used
45
  - `cvrplib/tai75a.vrp`: CVRP instance tai75a (75 customers) with best known solution 1618.36.
46
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
47
  Origin: published benchmark instance from Rochat and Taillard (1995), as served by CVRPLIB; recorded tool_fixture
48
- Reference result: best known total cost plus reference routes (proven optimal)
49
- Reference source: CVRPLIB BKS file tai75a.sol (.sol line 11 "Cost 1618.36" (COMMENT line 3 of tai75a.vrp: 1618.36)).
50
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in tai75a.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 1618.36 plus 0.005, the published best known solution (proven optimal, so a lower total means a checker error)
51
  Companions (not included): cvrplib/tai75a.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
52
  Dependencies: no upstream code used
53
  - `cvrplib/Golden_1.vrp`: CVRP instance Golden_1 (240 customers) with best known solution 5623.47.
54
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
55
  Origin: published benchmark instance from Golden, Wasil, Kelly and Chao (1998), as served by CVRPLIB; recorded tool_fixture
56
- Reference result: best known total cost plus reference routes (upper bound, not proven optimal)
57
- Reference source: CVRPLIB BKS file Golden_1.sol (.sol line 10 "Cost 5623.47"; the COMMENT line 2 of Golden_1.vrp (5627.54) is an older, worse value).
58
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Golden_1.vrp; each route length (depot to depot, no service times in this file) must not exceed DISTANCE 650.00000; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 5623.47 plus 0.005, the published best known solution
59
  Companions (not included): cvrplib/Golden_1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
60
  Dependencies: no upstream code used
61
  - `cvrplib/C101.txt`: VRPTW instance C101 with best known solution 827.3.
62
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
63
  Origin: published benchmark instance (Solomon (1987) set C1, 100 customers) as served by CVRPLIB; recorded tool_fixture
64
- Reference result: best known total distance plus reference routes (proven optimal)
65
- Reference source: CVRPLIB BKS file C101.sol (.sol line 11 "Cost 827.3").
66
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when every customer is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed), lasts SERVICE TIME, and the vehicle returns by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292: "The objective is to minimize only the total distance"); sum in tenths and pass when feasible and the total is at most 827.3, the published best known value (proven optimal)
67
  Companions (not included): cvrplib/C101.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/vrp-with-time-windows
68
  Dependencies: no upstream code used
69
  - `cvrplib/R1_2_1.txt`: VRPTW instance R1_2_1 with best known solution 4667.2.
70
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
71
  Origin: published benchmark instance (Gehring and Homberger (1999) set R1, 200 customers) as served by CVRPLIB; recorded tool_fixture
72
- Reference result: best known total distance plus reference routes (proven optimal)
73
- Reference source: CVRPLIB BKS file R1_2_1.sol (.sol line 24 "Cost 4667.2").
74
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when every customer is on exactly one route, at most 50 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed), lasts SERVICE TIME, and the vehicle returns by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292: "The objective is to minimize only the total distance"); sum in tenths and pass when feasible and the total is at most 4667.2, the published best known value (proven optimal)
75
  Companions (not included): cvrplib/R1_2_1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/vrp-with-time-windows
76
  Dependencies: no upstream code used
77
  - `cvrplib/RC2_10_1.txt`: VRPTW instance RC2_10_1 with best known solution 28122.6.
78
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
79
  Origin: published benchmark instance (Gehring and Homberger (1999) set RC2, 1000 customers) as served by CVRPLIB; recorded tool_fixture
80
- Reference result: best known total distance plus reference routes (upper bound)
81
- Reference source: CVRPLIB BKS file RC2_10_1.sol (.sol line 30 "Cost 28122.6").
82
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when every customer is on exactly one route, at most 250 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed), lasts SERVICE TIME, and the vehicle returns by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292: "The objective is to minimize only the total distance"); sum in tenths and pass when feasible and the total is at most 28122.6, the published best known value
83
  Companions (not included): cvrplib/RC2_10_1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/vrp-with-time-windows
84
  Dependencies: no upstream code used
85
  - `cvrplib/Li_21.vrp`: CVRP instance Li_21 (560 customers) with best known solution 16212.82548.
86
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
87
  Origin: published benchmark instance from Li, Golden and Wasil (2005), as served by CVRPLIB; recorded tool_fixture
88
- Reference result: best known total cost plus reference routes (upper bound, not proven optimal)
89
- Reference source: CVRPLIB BKS file Li_21.sol (.sol line 11 "Cost 16212.82548" (COMMENT line 2 of Li_21.vrp: 16212.74, an older rounded value; the index shows 16212.83)).
90
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Li_21.vrp; each route length (depot to depot, no service times in this file) must not exceed DISTANCE 1800; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 16212.82548 plus 0.000005, the published best known solution
91
  Companions (not included): cvrplib/Li_21.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
92
  Dependencies: no upstream code used
93
  - `cvrplib/Leuven1.vrp`: CVRP instance Leuven1 (3000 customers) with best known solution 192848.
94
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
95
  Origin: published benchmark instance from Arnold, Gendreau and Sorensen (2017), as served by CVRPLIB; recorded tool_fixture
96
- Reference result: best known total cost plus reference routes (upper bound, not proven optimal)
97
- Reference source: CVRPLIB BKS file Leuven1.sol (.sol line 204 "Cost 192848").
98
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Leuven1.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 192848, the published best known solution
99
  Companions (not included): cvrplib/Leuven1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
100
  Dependencies: no upstream code used
101
  - `cvrplib/Loggi-n401-k23.vrp`: CVRP instance Loggi-n401-k23 (400 customers) with best known solution 336903.
102
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
103
  Origin: published benchmark instance from DIMACS 2021 CVRP set (Loggi, Belem), as served by CVRPLIB (derived from real delivery data by the contributor, per COMMENT line 2); recorded tool_fixture
104
- Reference result: best known total cost plus reference routes (upper bound, not proven optimal)
105
- Reference source: CVRPLIB BKS file Loggi-n401-k23.sol (.sol line 24 "Cost 336903").
106
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Loggi-n401-k23.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, edge weights read from EDGE_WEIGHT_SECTION (EXPLICIT, LOWER_ROW), integers; pass when the routes are feasible and the recomputed integer total is at most 336903, the published best known solution
107
  Companions (not included): cvrplib/Loggi-n401-k23.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
108
  Dependencies: no upstream code used
109
  - `cvrplib/ORTEC-n242-k12.vrp`: CVRP instance ORTEC-n242-k12 (241 customers) with best known solution 123750.
110
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
111
  Origin: published benchmark instance from DIMACS 2021 CVRP set (ORTEC grocery delivery), as served by CVRPLIB (derived from real delivery data by the contributor, per COMMENT line 2); recorded tool_fixture
112
- Reference result: best known total cost plus reference routes (upper bound, not proven optimal)
113
- Reference source: CVRPLIB BKS file ORTEC-n242-k12.sol (.sol line 13 "Cost 123750").
114
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in ORTEC-n242-k12.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, edge weights read from EDGE_WEIGHT_SECTION (EXPLICIT, LOWER_ROW), integers; pass when the routes are feasible and the recomputed integer total is at most 123750, the published best known solution
115
  Companions (not included): cvrplib/ORTEC-n242-k12.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
116
  Dependencies: no upstream code used
117
  - `cvrplib/XL-n1048-k237.vrp`: CVRP instance XL-n1048-k237 (1047 customers) with best known solution 380092.
118
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
119
  Origin: published benchmark instance from CVRPLib BKS Challenge XL set, as served by CVRPLIB; recorded tool_fixture
120
- Reference result: best known total cost plus reference routes (upper bound, not proven optimal)
121
- Reference source: CVRPLIB BKS file XL-n1048-k237.sol (.sol line 239 "Cost 380092").
122
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in XL-n1048-k237.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 380092, the published best known solution
123
  Companions (not included): cvrplib/XL-n1048-k237.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
124
  Dependencies: no upstream code used
125
  - `cvrplib/B-n31-k5.vrp`: CVRP instance B-n31-k5 with optimal solution 672.
126
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
127
  Origin: published benchmark instance (Augerat et al. (1995) set B) as served by CVRPLIB; recorded tool_fixture
128
- Reference result: optimal total distance plus reference routes
129
- Reference source: CVRPLIB BKS file B-n31-k5.sol (line 6 "Cost 672", 5 routes), proven optimal per the CVRPLIB index.
130
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 30 customers is on exactly one route and each route load is at most Q = 100 (DEMAND_SECTION of B-n31-k5.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 672, the published value (proven optimal, so a lower total means a checker error)
131
  Companions (not included): cvrplib/B-n31-k5.sol
132
  Dependencies: no upstream code used
133
  - `cvrplib/E-n51-k5.vrp`: CVRP instance E-n51-k5 with optimal solution 521.
134
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
135
  Origin: published benchmark instance (Christofides and Eilon (1969) set E) as served by CVRPLIB; recorded tool_fixture
136
- Reference result: optimal total distance plus reference routes
137
- Reference source: CVRPLIB BKS file E-n51-k5.sol (line 6 "Cost 521", 5 routes), proven optimal per the CVRPLIB index.
138
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 50 customers is on exactly one route and each route load is at most Q = 160 (DEMAND_SECTION of E-n51-k5.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 521, the published value (proven optimal, so a lower total means a checker error)
139
  Companions (not included): cvrplib/E-n51-k5.sol
140
  Dependencies: no upstream code used
141
  - `cvrplib/F-n45-k4.vrp`: CVRP instance F-n45-k4 with optimal solution 724.
142
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
143
  Origin: published benchmark instance (Fisher (1994) set F) as served by CVRPLIB; recorded tool_fixture
144
- Reference result: optimal total distance plus reference routes
145
- Reference source: CVRPLIB BKS file F-n45-k4.sol (line 5 "Cost 724", 4 routes), proven optimal per the CVRPLIB index.
146
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 44 customers is on exactly one route and each route load is at most Q = 2010 (DEMAND_SECTION of F-n45-k4.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 724, the published value (proven optimal, so a lower total means a checker error)
147
  Companions (not included): cvrplib/F-n45-k4.sol
148
  Dependencies: no upstream code used
149
  - `cvrplib/M-n101-k10.vrp`: CVRP instance M-n101-k10 with optimal solution 820.
150
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
151
  Origin: published benchmark instance (Christofides, Mingozzi and Toth (1979) set M) as served by CVRPLIB; recorded tool_fixture
152
- Reference result: optimal total distance plus reference routes
153
- Reference source: CVRPLIB BKS file M-n101-k10.sol (line 11 "Cost 820", 10 routes), proven optimal per the CVRPLIB index.
154
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 100 customers is on exactly one route and each route load is at most Q = 200 (DEMAND_SECTION of M-n101-k10.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 820, the published value (proven optimal, so a lower total means a checker error)
155
  Companions (not included): cvrplib/M-n101-k10.sol
156
  Dependencies: no upstream code used
157
  - `cvrplib/P-n16-k8.vrp`: CVRP instance P-n16-k8 with optimal solution 450.
158
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
159
  Origin: published benchmark instance (Augerat et al. (1995) set P) as served by CVRPLIB; recorded tool_fixture
160
- Reference result: optimal total distance plus reference routes
161
- Reference source: CVRPLIB BKS file P-n16-k8.sol (line 9 "Cost 450", 8 routes), proven optimal per the CVRPLIB index.
162
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 15 customers is on exactly one route and each route load is at most Q = 35 (DEMAND_SECTION of P-n16-k8.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 450, the published value (proven optimal, so a lower total means a checker error)
163
  Companions (not included): cvrplib/P-n16-k8.sol
164
  Dependencies: no upstream code used
165
  - `cvrplib/tai385.vrp`: CVRP instance tai385 with best known solution 24366.41.
166
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
167
  Origin: published benchmark instance (Rochat and Taillard (1995), tai385) as served by CVRPLIB; recorded tool_fixture
168
- Reference result: best known total distance plus reference routes (not proven optimal)
169
- Reference source: CVRPLIB BKS file tai385.sol (line 48 "Cost 24366.41339", 47 routes) and the index UB 24366.41.
170
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 385 customers is on exactly one route and each route load is at most Q = 65; no route length limit in this file and no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 24366.41 plus 0.005, the best known value (not proven optimal, a lower total also passes)
171
  Companions (not included): cvrplib/tai385.sol
172
  Dependencies: no upstream code used
173
  - `cvrplib/Golden_9.vrp`: CVRP instance Golden_9 with best known solution 579.70.
174
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
175
  Origin: published benchmark instance (Golden, Wasil, Kelly and Chao (1998), Golden_9) as served by CVRPLIB; recorded tool_fixture
176
- Reference result: best known total distance plus reference routes (not proven optimal)
177
- Reference source: CVRPLIB BKS file Golden_9.sol (line 15 "Cost 579.702026", 14 routes) and the index UB 579.70.
178
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 255 customers is on exactly one route and each route load is at most Q = 1000; no route length limit in this file and no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 579.70 plus 0.005, the best known value (not proven optimal, a lower total also passes)
179
  Companions (not included): cvrplib/Golden_9.sol
180
  Dependencies: no upstream code used
181
  - `cvrplib/R101.txt`: VRPTW instance R101 with optimal solution 1637.7.
182
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
183
  Origin: published benchmark instance (Solomon (1987) set R1, 100 customers) as served by CVRPLIB; recorded tool_fixture
184
- Reference result: optimal total distance plus reference routes
185
- Reference source: CVRPLIB BKS file R101.sol (line 21 "Cost 1637.7").
186
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1637.7, the published value (proven optimal)
187
  Companions (not included): cvrplib/R101.sol
188
  Dependencies: no upstream code used
189
  - `cvrplib/RC101.txt`: VRPTW instance RC101 with optimal solution 1619.8.
190
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
191
  Origin: published benchmark instance (Solomon (1987) set RC1, 100 customers) as served by CVRPLIB; recorded tool_fixture
192
- Reference result: optimal total distance plus reference routes
193
- Reference source: CVRPLIB BKS file RC101.sol (line 16 "Cost 1619.8").
194
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1619.8, the published value (proven optimal)
195
  Companions (not included): cvrplib/RC101.sol
196
  Dependencies: no upstream code used
197
  - `cvrplib/C201.txt`: VRPTW instance C201 with optimal solution 589.1.
198
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
199
  Origin: published benchmark instance (Solomon (1987) set C2, 100 customers) as served by CVRPLIB; recorded tool_fixture
200
- Reference result: optimal total distance plus reference routes
201
- Reference source: CVRPLIB BKS file C201.sol (line 4 "Cost 589.1").
202
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 700, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 589.1, the published value (proven optimal)
203
  Companions (not included): cvrplib/C201.sol
204
  Dependencies: no upstream code used
205
  - `cvrplib/C1_4_1.txt`: VRPTW instance C1_4_1 with optimal solution 7138.8.
206
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
207
  Origin: published benchmark instance (Gehring and Homberger (1999) set C1, 400 customers) as served by CVRPLIB; recorded tool_fixture
208
- Reference result: optimal total distance plus reference routes
209
- Reference source: CVRPLIB BKS file C1_4_1.sol (line 41 "Cost 7138.8").
210
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 400 customers is on exactly one route, at most 100 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 7138.8, the published value (proven optimal)
211
  Companions (not included): cvrplib/C1_4_1.sol
212
  Dependencies: no upstream code used
213
  - `cvrplib/R2_6_1.txt`: VRPTW instance R2_6_1 with best known solution 15145.3.
214
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
215
  Origin: published benchmark instance (Gehring and Homberger (1999) set R2, 600 customers) as served by CVRPLIB; recorded tool_fixture
216
- Reference result: best known (not proven optimal) total distance plus reference routes
217
- Reference source: CVRPLIB BKS file R2_6_1.sol (line 28 "Cost 15145.3").
218
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 600 customers is on exactly one route, at most 150 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 15145.3, the published value (best known, not proven optimal; a lower total also passes)
219
  Companions (not included): cvrplib/R2_6_1.sol
220
  Dependencies: no upstream code used
221
  - `cvrplib/E-n101-k8.vrp`: CVRP instance E-n101-k8 with optimal solution 815.
222
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
223
  Origin: published benchmark instance (Christofides and Eilon (1969), set E) as served by CVRPLIB; recorded tool_fixture
224
- Reference result: optimal total distance plus reference routes
225
- Reference source: CVRPLIB BKS file E-n101-k8.sol (line 9 "Cost 815", 8 routes), proven optimal per the CVRPLIB index (UB 815.00).
226
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 100 customers is on exactly one route and each route load is at most Q = 200 (DEMAND_SECTION of E-n101-k8.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 815, the published value (proven optimal, so a lower total means a checker error)
227
  Companions (not included): cvrplib/E-n101-k8.sol
228
  Dependencies: no upstream code used
229
  - `cvrplib/P-n101-k4.vrp`: CVRP instance P-n101-k4 with optimal solution 681.
230
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
231
  Origin: published benchmark instance (Augerat et al. (1995), set P) as served by CVRPLIB; recorded tool_fixture
232
- Reference result: optimal total distance plus reference routes
233
- Reference source: CVRPLIB BKS file P-n101-k4.sol (line 5 "Cost 681", 4 routes), proven optimal per the CVRPLIB index (UB 681.00).
234
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 100 customers is on exactly one route and each route load is at most Q = 400 (DEMAND_SECTION of P-n101-k4.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 681, the published value (proven optimal, so a lower total means a checker error)
235
  Companions (not included): cvrplib/P-n101-k4.sol
236
  Dependencies: no upstream code used
237
  - `cvrplib/F-n135-k7.vrp`: CVRP instance F-n135-k7 with optimal solution 1162.
238
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
239
  Origin: published benchmark instance (Fisher (1994), set F) as served by CVRPLIB; recorded tool_fixture
240
- Reference result: optimal total distance plus reference routes
241
- Reference source: CVRPLIB BKS file F-n135-k7.sol (line 8 "Cost 1162", 7 routes), proven optimal per the CVRPLIB index (UB 1162.00).
242
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 134 customers is on exactly one route and each route load is at most Q = 2210 (DEMAND_SECTION of F-n135-k7.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 1162, the published value (proven optimal, so a lower total means a checker error)
243
  Companions (not included): cvrplib/F-n135-k7.sol
244
  Dependencies: no upstream code used
245
  - `cvrplib/A-n80-k10.vrp`: CVRP instance A-n80-k10 with optimal solution 1763.
246
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
247
  Origin: published benchmark instance (Augerat et al. (1995), set A) as served by CVRPLIB; recorded tool_fixture
248
- Reference result: optimal total distance plus reference routes
249
- Reference source: CVRPLIB BKS file A-n80-k10.sol (line 11 "Cost 1763", 10 routes), proven optimal per the CVRPLIB index (UB 1763.00).
250
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 79 customers is on exactly one route and each route load is at most Q = 100 (DEMAND_SECTION of A-n80-k10.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 1763, the published value (proven optimal, so a lower total means a checker error)
251
  Companions (not included): cvrplib/A-n80-k10.sol
252
  Dependencies: no upstream code used
253
  - `cvrplib/X-n106-k14.vrp`: CVRP instance X-n106-k14 with optimal solution 26362.
254
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance; this file has CRLF line ends and tab separated fields as served upstream
255
  Origin: published benchmark instance (Uchoa, Pecin, Pessoa, Poggi, Subramanian and Vidal (2017), set X) as served by CVRPLIB; recorded tool_fixture
256
- Reference result: optimal total distance plus reference routes
257
- Reference source: CVRPLIB BKS file X-n106-k14.sol (line 15 "Cost 26362", 14 routes), proven optimal per the CVRPLIB index (UB 26362.00).
258
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 105 customers is on exactly one route and each route load is at most Q = 600 (DEMAND_SECTION of X-n106-k14.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 26362, the published value (proven optimal, so a lower total means a checker error)
259
  Companions (not included): cvrplib/X-n106-k14.sol
260
  Dependencies: no upstream code used
261
  - `cvrplib/CMT4.vrp`: CVRP instance CMT4 with optimal solution 1028.42.
262
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
263
  Origin: published benchmark instance (Christofides, Mingozzi and Toth (1979), problem 4) as served by CVRPLIB; recorded tool_fixture
264
- Reference result: optimal total distance plus reference routes
265
- Reference source: CVRPLIB BKS file CMT4.sol (line 13 "Cost 1028.42", 12 routes), proven optimal per the CVRPLIB index (UB 1028.42).
266
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 150 customers is on exactly one route and each route load is at most Q = 200; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 1028.42 plus 0.005, the proven optimal value
267
  Companions (not included): cvrplib/CMT4.sol
268
  Dependencies: no upstream code used
269
  - `cvrplib/CMT5.vrp`: CVRP instance CMT5 with optimal solution 1291.29.
270
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
271
  Origin: published benchmark instance (Christofides, Mingozzi and Toth (1979), problem 5) as served by CVRPLIB; recorded tool_fixture
272
- Reference result: optimal total distance plus reference routes
273
- Reference source: CVRPLIB BKS file CMT5.sol (line 17 "Cost 1291.289144", 16 routes), proven optimal per the CVRPLIB index (UB 1291.29).
274
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 199 customers is on exactly one route and each route load is at most Q = 200; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 1291.29 plus 0.005, the proven optimal value
275
  Companions (not included): cvrplib/CMT5.sol
276
  Dependencies: no upstream code used
277
  - `cvrplib/Golden_13.vrp`: CVRP instance Golden_13 with optimal solution 857.19.
278
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
279
  Origin: published benchmark instance (Golden, Wasil, Kelly and Chao (1998), problem 13) as served by CVRPLIB; recorded tool_fixture
280
- Reference result: optimal total distance plus reference routes
281
- Reference source: CVRPLIB BKS file Golden_13.sol (line 27 "Cost 857.189", 26 routes), proven optimal per the CVRPLIB index (UB 857.19).
282
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 252 customers is on exactly one route and each route load is at most Q = 1000; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 857.19 plus 0.005, the proven optimal value
283
  Companions (not included): cvrplib/Golden_13.sol
284
  Dependencies: no upstream code used
285
  - `cvrplib/Golden_17.vrp`: CVRP instance Golden_17 with optimal solution 707.76.
286
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
287
  Origin: published benchmark instance (Golden, Wasil, Kelly and Chao (1998), problem 17) as served by CVRPLIB; recorded tool_fixture
288
- Reference result: optimal total distance plus reference routes
289
- Reference source: CVRPLIB BKS file Golden_17.sol (line 23 "Cost 707.756", 22 routes), proven optimal per the CVRPLIB index (UB 707.76).
290
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 240 customers is on exactly one route and each route load is at most Q = 200; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 707.76 plus 0.005, the proven optimal value
291
  Companions (not included): cvrplib/Golden_17.sol
292
  Dependencies: no upstream code used
293
  - `cvrplib/R201.txt`: VRPTW instance R201 with optimal solution 1143.2.
294
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance; this file has CRLF line ends as served upstream
295
  Origin: published benchmark instance (Solomon (1987), set R2, 100 customers) as served by CVRPLIB; recorded tool_fixture
296
- Reference result: optimal total distance plus reference routes
297
- Reference source: CVRPLIB BKS file R201.sol (line 9 "Cost 1143.2").
298
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1143.2, the published value (proven optimal)
299
  Companions (not included): cvrplib/R201.sol
300
  Dependencies: no upstream code used
301
  - `cvrplib/RC201.txt`: VRPTW instance RC201 with optimal solution 1261.8.
302
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance; this file has CRLF line ends as served upstream
303
  Origin: published benchmark instance (Solomon (1987), set RC2, 100 customers) as served by CVRPLIB; recorded tool_fixture
304
- Reference result: optimal total distance plus reference routes
305
- Reference source: CVRPLIB BKS file RC201.sol (line 10 "Cost 1261.8").
306
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1261.8, the published value (proven optimal)
307
  Companions (not included): cvrplib/RC201.sol
308
  Dependencies: no upstream code used
309
  - `cvrplib/C105.txt`: VRPTW instance C105 with optimal solution 827.3.
310
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance; this file has CRLF line ends as served upstream
311
  Origin: published benchmark instance (Solomon (1987), set C1, 100 customers) as served by CVRPLIB; recorded tool_fixture
312
- Reference result: optimal total distance plus reference routes
313
- Reference source: CVRPLIB BKS file C105.sol (line 11 "Cost 827.3").
314
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 827.3, the published value (proven optimal)
315
  Companions (not included): cvrplib/C105.sol
316
  Dependencies: no upstream code used
317
  - `cvrplib/C2_2_1.txt`: VRPTW instance C2_2_1 with optimal solution 1922.1.
318
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
319
  Origin: published benchmark instance (Gehring and Homberger (1999), set C2, 200 customers) as served by CVRPLIB; recorded tool_fixture
320
- Reference result: optimal total distance plus reference routes
321
- Reference source: CVRPLIB BKS file C2_2_1.sol (line 8 "Cost 1922.1").
322
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 200 customers is on exactly one route, at most 50 routes, each route load at most 700, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1922.1, the published value (proven optimal)
323
  Companions (not included): cvrplib/C2_2_1.sol
324
  Dependencies: no upstream code used
325
  - `cvrplib/RC1_2_1.txt`: VRPTW instance RC1_2_1 with optimal solution 3516.9.
326
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
327
  Origin: published benchmark instance (Gehring and Homberger (1999), set RC1, 200 customers) as served by CVRPLIB; recorded tool_fixture
328
- Reference result: optimal total distance plus reference routes
329
- Reference source: CVRPLIB BKS file RC1_2_1.sol (line 21 "Cost 3516.9").
330
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 200 customers is on exactly one route, at most 50 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 3516.9, the published value (proven optimal)
331
  Companions (not included): cvrplib/RC1_2_1.sol
332
  Dependencies: no upstream code used
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 39 of 39 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Tools are named in the check written per file in the notes.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `cvrplib/A-n32-k5.vrp`: CVRP instance A-n32-k5 (31 customers) with best known solution 784.
20
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
21
  Origin: published benchmark instance from Augerat et al. (1995) set A, as served by CVRPLIB; recorded tool_fixture
 
 
22
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in A-n32-k5.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 784, the published best known solution (proven optimal, so a lower total means a checker error)
23
  Companions (not included): cvrplib/A-n32-k5.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
24
  Dependencies: no upstream code used
25
  - `cvrplib/X-n101-k25.vrp`: CVRP instance X-n101-k25 (100 customers) with best known solution 27591.
26
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
27
  Origin: published benchmark instance from Uchoa et al. (2014) set X, as served by CVRPLIB; recorded tool_fixture
 
 
28
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in X-n101-k25.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 27591, the published best known solution (proven optimal, so a lower total means a checker error)
29
  Companions (not included): cvrplib/X-n101-k25.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
30
  Dependencies: no upstream code used
31
  - `cvrplib/CMT1.vrp`: CVRP instance CMT1 (50 customers) with best known solution 524.611.
32
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
33
  Origin: published benchmark instance from Christofides, Mingozzi and Toth (1979), as served by CVRPLIB; recorded tool_fixture
 
 
34
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in CMT1.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 524.611 plus 0.0005, the published best known solution (proven optimal, so a lower total means a checker error)
35
  Companions (not included): cvrplib/CMT1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
36
  Dependencies: no upstream code used
37
  - `cvrplib/tai75a.vrp`: CVRP instance tai75a (75 customers) with best known solution 1618.36.
38
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
39
  Origin: published benchmark instance from Rochat and Taillard (1995), as served by CVRPLIB; recorded tool_fixture
 
 
40
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in tai75a.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 1618.36 plus 0.005, the published best known solution (proven optimal, so a lower total means a checker error)
41
  Companions (not included): cvrplib/tai75a.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
42
  Dependencies: no upstream code used
43
  - `cvrplib/Golden_1.vrp`: CVRP instance Golden_1 (240 customers) with best known solution 5623.47.
44
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
45
  Origin: published benchmark instance from Golden, Wasil, Kelly and Chao (1998), as served by CVRPLIB; recorded tool_fixture
 
 
46
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Golden_1.vrp; each route length (depot to depot, no service times in this file) must not exceed DISTANCE 650.00000; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 5623.47 plus 0.005, the published best known solution
47
  Companions (not included): cvrplib/Golden_1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
48
  Dependencies: no upstream code used
49
  - `cvrplib/C101.txt`: VRPTW instance C101 with best known solution 827.3.
50
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
51
  Origin: published benchmark instance (Solomon (1987) set C1, 100 customers) as served by CVRPLIB; recorded tool_fixture
 
 
52
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when every customer is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed), lasts SERVICE TIME, and the vehicle returns by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292: "The objective is to minimize only the total distance"); sum in tenths and pass when feasible and the total is at most 827.3, the published best known value (proven optimal)
53
  Companions (not included): cvrplib/C101.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/vrp-with-time-windows
54
  Dependencies: no upstream code used
55
  - `cvrplib/R1_2_1.txt`: VRPTW instance R1_2_1 with best known solution 4667.2.
56
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
57
  Origin: published benchmark instance (Gehring and Homberger (1999) set R1, 200 customers) as served by CVRPLIB; recorded tool_fixture
 
 
58
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when every customer is on exactly one route, at most 50 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed), lasts SERVICE TIME, and the vehicle returns by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292: "The objective is to minimize only the total distance"); sum in tenths and pass when feasible and the total is at most 4667.2, the published best known value (proven optimal)
59
  Companions (not included): cvrplib/R1_2_1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/vrp-with-time-windows
60
  Dependencies: no upstream code used
61
  - `cvrplib/RC2_10_1.txt`: VRPTW instance RC2_10_1 with best known solution 28122.6.
62
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
63
  Origin: published benchmark instance (Gehring and Homberger (1999) set RC2, 1000 customers) as served by CVRPLIB; recorded tool_fixture
 
 
64
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when every customer is on exactly one route, at most 250 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed), lasts SERVICE TIME, and the vehicle returns by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292: "The objective is to minimize only the total distance"); sum in tenths and pass when feasible and the total is at most 28122.6, the published best known value
65
  Companions (not included): cvrplib/RC2_10_1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/vrp-with-time-windows
66
  Dependencies: no upstream code used
67
  - `cvrplib/Li_21.vrp`: CVRP instance Li_21 (560 customers) with best known solution 16212.82548.
68
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
69
  Origin: published benchmark instance from Li, Golden and Wasil (2005), as served by CVRPLIB; recorded tool_fixture
 
 
70
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Li_21.vrp; each route length (depot to depot, no service times in this file) must not exceed DISTANCE 1800; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, unrounded Euclidean distance from NODE_COORD_SECTION (real numbers); pass when the routes are feasible and the recomputed real total is at most 16212.82548 plus 0.000005, the published best known solution
71
  Companions (not included): cvrplib/Li_21.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
72
  Dependencies: no upstream code used
73
  - `cvrplib/Leuven1.vrp`: CVRP instance Leuven1 (3000 customers) with best known solution 192848.
74
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
75
  Origin: published benchmark instance from Arnold, Gendreau and Sorensen (2017), as served by CVRPLIB; recorded tool_fixture
 
 
76
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Leuven1.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 192848, the published best known solution
77
  Companions (not included): cvrplib/Leuven1.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
78
  Dependencies: no upstream code used
79
  - `cvrplib/Loggi-n401-k23.vrp`: CVRP instance Loggi-n401-k23 (400 customers) with best known solution 336903.
80
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
81
  Origin: published benchmark instance from DIMACS 2021 CVRP set (Loggi, Belem), as served by CVRPLIB (derived from real delivery data by the contributor, per COMMENT line 2); recorded tool_fixture
 
 
82
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in Loggi-n401-k23.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, edge weights read from EDGE_WEIGHT_SECTION (EXPLICIT, LOWER_ROW), integers; pass when the routes are feasible and the recomputed integer total is at most 336903, the published best known solution
83
  Companions (not included): cvrplib/Loggi-n401-k23.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
84
  Dependencies: no upstream code used
85
  - `cvrplib/ORTEC-n242-k12.vrp`: CVRP instance ORTEC-n242-k12 (241 customers) with best known solution 123750.
86
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
87
  Origin: published benchmark instance from DIMACS 2021 CVRP set (ORTEC grocery delivery), as served by CVRPLIB (derived from real delivery data by the contributor, per COMMENT line 2); recorded tool_fixture
 
 
88
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in ORTEC-n242-k12.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, edge weights read from EDGE_WEIGHT_SECTION (EXPLICIT, LOWER_ROW), integers; pass when the routes are feasible and the recomputed integer total is at most 123750, the published best known solution
89
  Companions (not included): cvrplib/ORTEC-n242-k12.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
90
  Dependencies: no upstream code used
91
  - `cvrplib/XL-n1048-k237.vrp`: CVRP instance XL-n1048-k237 (1047 customers) with best known solution 380092.
92
  Format: TSPLIB95-style CVRP file (NAME, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION or EDGE_WEIGHT_SECTION, DEMAND_SECTION, DEPOT_SECTION); one case = one instance; the answer is a set of routes and their total cost
93
  Origin: published benchmark instance from CVRPLib BKS Challenge XL set, as served by CVRPLIB; recorded tool_fixture
 
 
94
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when every customer is on exactly one route and each route load is at most the capacity Q given in XL-n1048-k237.vrp; no limit on the number of routes (capacitated-vrp line 264: "it will be assumed that there are no restrictions to the number of routes in a solution"); cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when the routes are feasible and the recomputed integer total is at most 380092, the published best known solution
95
  Companions (not included): cvrplib/XL-n1048-k237.sol; programs/implementation-challenges/vehicle-routing/implementation-challenge-vehicle-routing/capacitated-vrp
96
  Dependencies: no upstream code used
97
  - `cvrplib/B-n31-k5.vrp`: CVRP instance B-n31-k5 with optimal solution 672.
98
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
99
  Origin: published benchmark instance (Augerat et al. (1995) set B) as served by CVRPLIB; recorded tool_fixture
 
 
100
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 30 customers is on exactly one route and each route load is at most Q = 100 (DEMAND_SECTION of B-n31-k5.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 672, the published value (proven optimal, so a lower total means a checker error)
101
  Companions (not included): cvrplib/B-n31-k5.sol
102
  Dependencies: no upstream code used
103
  - `cvrplib/E-n51-k5.vrp`: CVRP instance E-n51-k5 with optimal solution 521.
104
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
105
  Origin: published benchmark instance (Christofides and Eilon (1969) set E) as served by CVRPLIB; recorded tool_fixture
 
 
106
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 50 customers is on exactly one route and each route load is at most Q = 160 (DEMAND_SECTION of E-n51-k5.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 521, the published value (proven optimal, so a lower total means a checker error)
107
  Companions (not included): cvrplib/E-n51-k5.sol
108
  Dependencies: no upstream code used
109
  - `cvrplib/F-n45-k4.vrp`: CVRP instance F-n45-k4 with optimal solution 724.
110
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
111
  Origin: published benchmark instance (Fisher (1994) set F) as served by CVRPLIB; recorded tool_fixture
 
 
112
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 44 customers is on exactly one route and each route load is at most Q = 2010 (DEMAND_SECTION of F-n45-k4.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 724, the published value (proven optimal, so a lower total means a checker error)
113
  Companions (not included): cvrplib/F-n45-k4.sol
114
  Dependencies: no upstream code used
115
  - `cvrplib/M-n101-k10.vrp`: CVRP instance M-n101-k10 with optimal solution 820.
116
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
117
  Origin: published benchmark instance (Christofides, Mingozzi and Toth (1979) set M) as served by CVRPLIB; recorded tool_fixture
 
 
118
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 100 customers is on exactly one route and each route load is at most Q = 200 (DEMAND_SECTION of M-n101-k10.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 820, the published value (proven optimal, so a lower total means a checker error)
119
  Companions (not included): cvrplib/M-n101-k10.sol
120
  Dependencies: no upstream code used
121
  - `cvrplib/P-n16-k8.vrp`: CVRP instance P-n16-k8 with optimal solution 450.
122
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
123
  Origin: published benchmark instance (Augerat et al. (1995) set P) as served by CVRPLIB; recorded tool_fixture
 
 
124
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 15 customers is on exactly one route and each route load is at most Q = 35 (DEMAND_SECTION of P-n16-k8.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 450, the published value (proven optimal, so a lower total means a checker error)
125
  Companions (not included): cvrplib/P-n16-k8.sol
126
  Dependencies: no upstream code used
127
  - `cvrplib/tai385.vrp`: CVRP instance tai385 with best known solution 24366.41.
128
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
129
  Origin: published benchmark instance (Rochat and Taillard (1995), tai385) as served by CVRPLIB; recorded tool_fixture
 
 
130
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 385 customers is on exactly one route and each route load is at most Q = 65; no route length limit in this file and no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 24366.41 plus 0.005, the best known value (not proven optimal, a lower total also passes)
131
  Companions (not included): cvrplib/tai385.sol
132
  Dependencies: no upstream code used
133
  - `cvrplib/Golden_9.vrp`: CVRP instance Golden_9 with best known solution 579.70.
134
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
135
  Origin: published benchmark instance (Golden, Wasil, Kelly and Chao (1998), Golden_9) as served by CVRPLIB; recorded tool_fixture
 
 
136
  Check: Program: read the agent routes (CVRPLIB .sol layout, customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 255 customers is on exactly one route and each route load is at most Q = 1000; no route length limit in this file and no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 579.70 plus 0.005, the best known value (not proven optimal, a lower total also passes)
137
  Companions (not included): cvrplib/Golden_9.sol
138
  Dependencies: no upstream code used
139
  - `cvrplib/R101.txt`: VRPTW instance R101 with optimal solution 1637.7.
140
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
141
  Origin: published benchmark instance (Solomon (1987) set R1, 100 customers) as served by CVRPLIB; recorded tool_fixture
 
 
142
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1637.7, the published value (proven optimal)
143
  Companions (not included): cvrplib/R101.sol
144
  Dependencies: no upstream code used
145
  - `cvrplib/RC101.txt`: VRPTW instance RC101 with optimal solution 1619.8.
146
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
147
  Origin: published benchmark instance (Solomon (1987) set RC1, 100 customers) as served by CVRPLIB; recorded tool_fixture
 
 
148
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1619.8, the published value (proven optimal)
149
  Companions (not included): cvrplib/RC101.sol
150
  Dependencies: no upstream code used
151
  - `cvrplib/C201.txt`: VRPTW instance C201 with optimal solution 589.1.
152
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
153
  Origin: published benchmark instance (Solomon (1987) set C2, 100 customers) as served by CVRPLIB; recorded tool_fixture
 
 
154
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 700, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 589.1, the published value (proven optimal)
155
  Companions (not included): cvrplib/C201.sol
156
  Dependencies: no upstream code used
157
  - `cvrplib/C1_4_1.txt`: VRPTW instance C1_4_1 with optimal solution 7138.8.
158
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
159
  Origin: published benchmark instance (Gehring and Homberger (1999) set C1, 400 customers) as served by CVRPLIB; recorded tool_fixture
 
 
160
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 400 customers is on exactly one route, at most 100 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 7138.8, the published value (proven optimal)
161
  Companions (not included): cvrplib/C1_4_1.sol
162
  Dependencies: no upstream code used
163
  - `cvrplib/R2_6_1.txt`: VRPTW instance R2_6_1 with best known solution 15145.3.
164
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
165
  Origin: published benchmark instance (Gehring and Homberger (1999) set R2, 600 customers) as served by CVRPLIB; recorded tool_fixture
 
 
166
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 600 customers is on exactly one route, at most 150 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 15145.3, the published value (best known, not proven optimal; a lower total also passes)
167
  Companions (not included): cvrplib/R2_6_1.sol
168
  Dependencies: no upstream code used
169
  - `cvrplib/E-n101-k8.vrp`: CVRP instance E-n101-k8 with optimal solution 815.
170
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
171
  Origin: published benchmark instance (Christofides and Eilon (1969), set E) as served by CVRPLIB; recorded tool_fixture
 
 
172
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 100 customers is on exactly one route and each route load is at most Q = 200 (DEMAND_SECTION of E-n101-k8.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 815, the published value (proven optimal, so a lower total means a checker error)
173
  Companions (not included): cvrplib/E-n101-k8.sol
174
  Dependencies: no upstream code used
175
  - `cvrplib/P-n101-k4.vrp`: CVRP instance P-n101-k4 with optimal solution 681.
176
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
177
  Origin: published benchmark instance (Augerat et al. (1995), set P) as served by CVRPLIB; recorded tool_fixture
 
 
178
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 100 customers is on exactly one route and each route load is at most Q = 400 (DEMAND_SECTION of P-n101-k4.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 681, the published value (proven optimal, so a lower total means a checker error)
179
  Companions (not included): cvrplib/P-n101-k4.sol
180
  Dependencies: no upstream code used
181
  - `cvrplib/F-n135-k7.vrp`: CVRP instance F-n135-k7 with optimal solution 1162.
182
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
183
  Origin: published benchmark instance (Fisher (1994), set F) as served by CVRPLIB; recorded tool_fixture
 
 
184
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 134 customers is on exactly one route and each route load is at most Q = 2210 (DEMAND_SECTION of F-n135-k7.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 1162, the published value (proven optimal, so a lower total means a checker error)
185
  Companions (not included): cvrplib/F-n135-k7.sol
186
  Dependencies: no upstream code used
187
  - `cvrplib/A-n80-k10.vrp`: CVRP instance A-n80-k10 with optimal solution 1763.
188
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
189
  Origin: published benchmark instance (Augerat et al. (1995), set A) as served by CVRPLIB; recorded tool_fixture
 
 
190
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 79 customers is on exactly one route and each route load is at most Q = 100 (DEMAND_SECTION of A-n80-k10.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 1763, the published value (proven optimal, so a lower total means a checker error)
191
  Companions (not included): cvrplib/A-n80-k10.sol
192
  Dependencies: no upstream code used
193
  - `cvrplib/X-n106-k14.vrp`: CVRP instance X-n106-k14 with optimal solution 26362.
194
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance; this file has CRLF line ends and tab separated fields as served upstream
195
  Origin: published benchmark instance (Uchoa, Pecin, Pessoa, Poggi, Subramanian and Vidal (2017), set X) as served by CVRPLIB; recorded tool_fixture
 
 
196
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 105 customers is on exactly one route and each route load is at most Q = 600 (DEMAND_SECTION of X-n106-k14.vrp); no limit on the number of routes; cost is the sum of edge lengths, Euclidean distance from NODE_COORD_SECTION rounded to the nearest integer per edge (TSPLIB EUC_2D); pass when feasible and the integer total is at most 26362, the published value (proven optimal, so a lower total means a checker error)
197
  Companions (not included): cvrplib/X-n106-k14.sol
198
  Dependencies: no upstream code used
199
  - `cvrplib/CMT4.vrp`: CVRP instance CMT4 with optimal solution 1028.42.
200
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
201
  Origin: published benchmark instance (Christofides, Mingozzi and Toth (1979), problem 4) as served by CVRPLIB; recorded tool_fixture
 
 
202
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 150 customers is on exactly one route and each route load is at most Q = 200; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 1028.42 plus 0.005, the proven optimal value
203
  Companions (not included): cvrplib/CMT4.sol
204
  Dependencies: no upstream code used
205
  - `cvrplib/CMT5.vrp`: CVRP instance CMT5 with optimal solution 1291.29.
206
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
207
  Origin: published benchmark instance (Christofides, Mingozzi and Toth (1979), problem 5) as served by CVRPLIB; recorded tool_fixture
 
 
208
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 199 customers is on exactly one route and each route load is at most Q = 200; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 1291.29 plus 0.005, the proven optimal value
209
  Companions (not included): cvrplib/CMT5.sol
210
  Dependencies: no upstream code used
211
  - `cvrplib/Golden_13.vrp`: CVRP instance Golden_13 with optimal solution 857.19.
212
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
213
  Origin: published benchmark instance (Golden, Wasil, Kelly and Chao (1998), problem 13) as served by CVRPLIB; recorded tool_fixture
 
 
214
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 252 customers is on exactly one route and each route load is at most Q = 1000; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 857.19 plus 0.005, the proven optimal value
215
  Companions (not included): cvrplib/Golden_13.sol
216
  Dependencies: no upstream code used
217
  - `cvrplib/Golden_17.vrp`: CVRP instance Golden_17 with optimal solution 707.76.
218
  Format: TSPLIB95 CVRP text: NAME, COMMENT, TYPE, DIMENSION, EDGE_WEIGHT_TYPE, CAPACITY, NODE_COORD_SECTION, DEMAND_SECTION, DEPOT_SECTION; one case = one instance; the answer is a set of routes and their total distance
219
  Origin: published benchmark instance (Golden, Wasil, Kelly and Chao (1998), problem 17) as served by CVRPLIB; recorded tool_fixture
 
 
220
  Check: Program: read the agent routes (CVRPLIB .sol layout "Route #k: ...", customers numbered node id minus 1, depot node 1 implicit at both ends); feasible when each of the 240 customers is on exactly one route and each route load is at most Q = 200; no route length limit and no service time in this file, no limit on the number of routes; cost is the sum of unrounded Euclidean edge lengths from NODE_COORD_SECTION; pass when feasible and the total is at most 707.76 plus 0.005, the proven optimal value
221
  Companions (not included): cvrplib/Golden_17.sol
222
  Dependencies: no upstream code used
223
  - `cvrplib/R201.txt`: VRPTW instance R201 with optimal solution 1143.2.
224
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance; this file has CRLF line ends as served upstream
225
  Origin: published benchmark instance (Solomon (1987), set R2, 100 customers) as served by CVRPLIB; recorded tool_fixture
 
 
226
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1143.2, the published value (proven optimal)
227
  Companions (not included): cvrplib/R201.sol
228
  Dependencies: no upstream code used
229
  - `cvrplib/RC201.txt`: VRPTW instance RC201 with optimal solution 1261.8.
230
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance; this file has CRLF line ends as served upstream
231
  Origin: published benchmark instance (Solomon (1987), set RC2, 100 customers) as served by CVRPLIB; recorded tool_fixture
 
 
232
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 1000, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1261.8, the published value (proven optimal)
233
  Companions (not included): cvrplib/RC201.sol
234
  Dependencies: no upstream code used
235
  - `cvrplib/C105.txt`: VRPTW instance C105 with optimal solution 827.3.
236
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance; this file has CRLF line ends as served upstream
237
  Origin: published benchmark instance (Solomon (1987), set C1, 100 customers) as served by CVRPLIB; recorded tool_fixture
 
 
238
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 100 customers is on exactly one route, at most 25 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 827.3, the published value (proven optimal)
239
  Companions (not included): cvrplib/C105.sol
240
  Dependencies: no upstream code used
241
  - `cvrplib/C2_2_1.txt`: VRPTW instance C2_2_1 with optimal solution 1922.1.
242
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
243
  Origin: published benchmark instance (Gehring and Homberger (1999), set C2, 200 customers) as served by CVRPLIB; recorded tool_fixture
 
 
244
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 200 customers is on exactly one route, at most 50 routes, each route load at most 700, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 1922.1, the published value (proven optimal)
245
  Companions (not included): cvrplib/C2_2_1.sol
246
  Dependencies: no upstream code used
247
  - `cvrplib/RC1_2_1.txt`: VRPTW instance RC1_2_1 with optimal solution 3516.9.
248
  Format: Solomon VRPTW text: name, VEHICLE NUMBER CAPACITY, then CUSTOMER rows CUST NO., XCOORD., YCOORD., DEMAND, READY TIME, DUE DATE, SERVICE TIME (row 0 is the depot); one case = one instance; the answer is routes and their total distance
249
  Origin: published benchmark instance (Gehring and Homberger (1999), set RC1, 200 customers) as served by CVRPLIB; recorded tool_fixture
 
 
250
  Check: Program: read the agent routes (CVRPLIB VRPTW .sol layout, customers by number, depot 0 implicit at both ends); feasible when each of the 200 customers is on exactly one route, at most 50 routes, each route load at most 200, service at each customer starts within its READY TIME and DUE DATE (waiting allowed) and lasts SERVICE TIME, and the vehicle is back by the depot due date; travel times equal distances, Euclidean truncated to 1 decimal per edge; objective is total distance only (vrp-with-time-windows line 292); sum in tenths and pass when feasible and the total is at most 3516.9, the published value (proven optimal)
251
  Companions (not included): cvrplib/RC1_2_1.sol
252
  Dependencies: no upstream code used
campaign_files/2026-10-g-solver-instances-2/dimacs-graph-coloring-instances/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Published chromatic number (optimal colouring size). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 6 distinct colours, the published chromatic number; Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 15 distinct colours, the published chromatic number; Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 10 distinct colours, the published chromatic number; Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 9 distinct colours, the published chromatic number.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,32 +19,24 @@ Files:
21
  - `COLOR/instances/myciel5.col`: DIMACS graph colouring instance myciel5.col: Mycielski graph, triangle free with chromatic number 6, 47 vertices, with its chromatic number.
22
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
23
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
24
- Reference result: published chromatic number (optimal colouring size)
25
- Reference source: COLOR/instances.html line 94: "myciel5.col </a> (47,236), 6", source MYC (Michael Trick); line 14 says "Each instance includes the information: (nodes, edges), optimal coloring, source." Published by the page author from the DIMACS second implementation challenge.
26
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 6 distinct colours, the published chromatic number
27
  Companions (not included): COLOR/instances.html
28
  Dependencies: not packed
29
  - `COLOR/instances/le450_15a.col`: DIMACS graph colouring instance le450_15a.col: Leighton graph with a planted 15-colouring, 450 vertices, with its chromatic number.
30
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
31
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
32
- Reference result: published chromatic number (optimal colouring size)
33
- Reference source: COLOR/instances.html line 46: "le450_15a.col </a> (450,8168), 15", source LEI (Craig Morgenstern); line 14 says "Each instance includes the information: (nodes, edges), optimal coloring, source." Published by the page author from the DIMACS second implementation challenge.
34
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 15 distinct colours, the published chromatic number
35
  Companions (not included): COLOR/instances.html
36
  Dependencies: not packed
37
  - `COLOR/instances/queen9_9.col`: DIMACS graph colouring instance queen9_9.col: 9x9 queen graph (two squares adjacent when a queen on one attacks the other), with its chromatic number.
38
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
39
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
40
- Reference result: published chromatic number (optimal colouring size)
41
- Reference source: COLOR/instances.html line 91: "queen9_9.col </a> (81, 2112), 10", source SGB (Stanford GraphBase); line 14 says "Each instance includes the information: (nodes, edges), optimal coloring, source." Published by the page author from the DIMACS second implementation challenge.
42
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 10 distinct colours, the published chromatic number
43
  Companions (not included): COLOR/instances.html
44
  Dependencies: not packed
45
  - `COLOR/instances/games120.col`: DIMACS graph colouring instance games120.col: graph of 120 US college football teams joined when they played in the 1990 season (vertices numbered, no team names in the file), with its chromatic number.
46
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
47
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
48
- Reference result: published chromatic number (optimal colouring size)
49
- Reference source: COLOR/instances.html line 73: "games120.col </a> (120,638), 9", source SGB (Stanford GraphBase); line 14 says "Each instance includes the information: (nodes, edges), optimal coloring, source." Published by the page author from the DIMACS second implementation challenge.
50
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 9 distinct colours, the published chromatic number
51
  Companions (not included): COLOR/instances.html
52
  Dependencies: not packed
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 6 distinct colours, the published chromatic number; Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 15 distinct colours, the published chromatic number; Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 10 distinct colours, the published chromatic number; Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 9 distinct colours, the published chromatic number.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `COLOR/instances/myciel5.col`: DIMACS graph colouring instance myciel5.col: Mycielski graph, triangle free with chromatic number 6, 47 vertices, with its chromatic number.
20
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
21
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
 
 
22
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 6 distinct colours, the published chromatic number
23
  Companions (not included): COLOR/instances.html
24
  Dependencies: not packed
25
  - `COLOR/instances/le450_15a.col`: DIMACS graph colouring instance le450_15a.col: Leighton graph with a planted 15-colouring, 450 vertices, with its chromatic number.
26
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
27
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
 
 
28
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 15 distinct colours, the published chromatic number
29
  Companions (not included): COLOR/instances.html
30
  Dependencies: not packed
31
  - `COLOR/instances/queen9_9.col`: DIMACS graph colouring instance queen9_9.col: 9x9 queen graph (two squares adjacent when a queen on one attacks the other), with its chromatic number.
32
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
33
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
 
 
34
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 10 distinct colours, the published chromatic number
35
  Companions (not included): COLOR/instances.html
36
  Dependencies: not packed
37
  - `COLOR/instances/games120.col`: DIMACS graph colouring instance games120.col: graph of 120 US college football teams joined when they played in the 1990 season (vertices numbered, no team names in the file), with its chromatic number.
38
  Format: DIMACS edge format: comment lines "c", one line "p edge n m", then lines "e u v" (1-based); one file is one case
39
  Origin: benchmark graph collected for the DIMACS second implementation challenge (1992 to 1993) and published with its chromatic number for testing colouring algorithms
 
 
40
  Check: Exact: the agent returns a colour for every vertex; the colouring must be proper (no edge with both ends of one colour, graph read as undirected) and use no more than 9 distinct colours, the published chromatic number
41
  Companions (not included): COLOR/instances.html
42
  Dependencies: not packed
campaign_files/2026-10-g-solver-instances/csplib/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Best known objective value (minimum host count 13), not a proven optimum for the statement as written. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: The agent returns a host set and a 6-period timetable; Correct when the timetable is feasible and the host count is <= 13: 13 is only the best known value, so a feasible solution with fewer hosts is also correct; optimality is never required or checked. Infeasible or more than 13 hosts is wrong.
14
 
15
  **Licence (SPDX):** CC-BY-4.0
@@ -21,8 +19,6 @@ Files:
21
  - `Problems/prob013/specification.md`: CSPLib prob013 Progressive Party Problem: the original 42-boat, 6-period yacht club instance; best known minimum number of host boats is 13.
22
  Format: Markdown: problem statement plus the boat table (boat number, capacity, crew size, 42 boats); companion results/basic_results.inline.md states the known optimum
23
  Origin: real yacht club rally timetabling problem (Smith et al. 1996) recorded in CSPLib
24
- Reference result: best known objective value (minimum host count 13), not a proven optimum for the statement as written.
25
- Reference source: Best known, not proven optimal for the CSPLib statement. Sources: (1) results/basic_results.inline.md line 1: "an optimal solution with 13 hosts was found by Barbara Smith, et al. ... The problem is trivially infeasible with just 12 hosts." (2) H.P. Williams's account (co-author) on J.P. Walser's page https://www.ps.uni-saarland.de/~walser/ppp/ppp.html (last updated 1997-03-10): the ILP team showed "quite easily, that a solution using only 12 Host boats was impossible"; Barbara Smith found 13 with ILOG Solver, "although this does not seem a viable method of PROVING optimality"; written up in Brailsford, Hubbard, Smith, Williams, Computers & OR 23(9):845-856, 1996, doi 10.1016/0305-0548(96)00001-9. (3) Smith, Brailsford, Hubbard, Williams, Leeds report 95.8 (March 1995), CP'95 LNCS 976 pp. 36-52 (doi 10.1007/3-540-60299-2_3), Constraints 1:119-138 (1996): full text paywalled, not read; the abstract only says a solution was found. (4) Kalvelagen 2002, https://amsterdamoptimization.com/pdf/ppp.pdf p.1: the real-world problem had extra conditions, boats 1 to 3 designated hosts and boats 40 to 42 guests; Appendix A fixes them and comments "it can be shows that 12 host boats is not enough to handle even a single period". Hence recorded as best known.
26
  Check: The agent returns a host set and a 6-period timetable; Correct when the timetable is feasible and the host count is <= 13: 13 is only the best known value, so a feasible solution with fewer hosts is also correct; optimality is never required or checked. Infeasible or more than 13 hosts is wrong.
27
  Companions (not included): Problems/prob013/results/basic_results.inline.md
28
  Dependencies: not packed
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: The agent returns a host set and a 6-period timetable; Correct when the timetable is feasible and the host count is <= 13: 13 is only the best known value, so a feasible solution with fewer hosts is also correct; optimality is never required or checked. Infeasible or more than 13 hosts is wrong.
12
 
13
  **Licence (SPDX):** CC-BY-4.0
 
19
  - `Problems/prob013/specification.md`: CSPLib prob013 Progressive Party Problem: the original 42-boat, 6-period yacht club instance; best known minimum number of host boats is 13.
20
  Format: Markdown: problem statement plus the boat table (boat number, capacity, crew size, 42 boats); companion results/basic_results.inline.md states the known optimum
21
  Origin: real yacht club rally timetabling problem (Smith et al. 1996) recorded in CSPLib
 
 
22
  Check: The agent returns a host set and a 6-period timetable; Correct when the timetable is feasible and the host count is <= 13: 13 is only the best known value, so a feasible solution with fewer hosts is also correct; optimality is never required or checked. Infeasible or more than 13 hosts is wrong.
23
  Companions (not included): Problems/prob013/results/basic_results.inline.md
24
  Dependencies: not packed
campaign_files/2026-10-gt-browser-tests-2/cpython-test-email-data/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: golden_output_files. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Python email package (message_from_file, Generator with maxheaderlen=0), or any MIME library that preserves the signed part; a script writes the re-flattened message to a file; Python email.mime (MIMEMultipart, MIMEText) and as_string, or mutt -H with a prepared draft; a script composes the message and writes it to a file
14
 
15
  **Licence (SPDX):** PSF-2.0
@@ -21,14 +19,10 @@ Files:
21
  - `Lib/test/test_email/data/msg_45.txt`: Re-serialize a PGP-signed multipart message without changing the signed part.
22
  Format: RFC 5322 message (.txt)
23
  Origin: CPython email test data (tool_fixture)
24
- Reference result: golden_output_files
25
- Reference source: the input file itself: TestIdempotent._idempotent (test_email.py lines 2857-2862) flattens with Generator maxheaderlen=0 and requires equality, test_message_signed_idempotent lines 2952-2954; TestSigned lines 5958-5978 require the first signed part bytes to be preserved
26
  Check: Final file state: the re-flattened message must equal msg_45.txt exactly, and the signed part must be byte-identical. No screenshot
27
  Tools: Python email package (message_from_file, Generator with maxheaderlen=0), or any MIME library that preserves the signed part; a script writes the re-flattened message to a file
28
  - `Lib/test/test_email/data/msg_21.txt`: Compose a two-part MIME message with a preamble and an epilogue from a spec.
29
  Format: RFC 5322 message (.txt)
30
  Origin: CPython email test data (tool_fixture)
31
- Reference result: golden_output_files
32
- Reference source: msg_21.txt is the expected output of test_epilogue (test_email.py lines 2638-2656): From aperson@dom.ain, To bperson@dom.ain, Subject Test, preamble MIME message, epilogue End of MIME message, parts One and Two, boundary BOUNDARY; also round-trip at lines 2912-2914
33
  Check: Final file state: the composed message must equal msg_21.txt exactly (assertEqual on as_string). No screenshot
34
  Tools: Python email.mime (MIMEMultipart, MIMEText) and as_string, or mutt -H with a prepared draft; a script composes the message and writes it to a file
 
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Python email package (message_from_file, Generator with maxheaderlen=0), or any MIME library that preserves the signed part; a script writes the re-flattened message to a file; Python email.mime (MIMEMultipart, MIMEText) and as_string, or mutt -H with a prepared draft; a script composes the message and writes it to a file
12
 
13
  **Licence (SPDX):** PSF-2.0
 
19
  - `Lib/test/test_email/data/msg_45.txt`: Re-serialize a PGP-signed multipart message without changing the signed part.
20
  Format: RFC 5322 message (.txt)
21
  Origin: CPython email test data (tool_fixture)
 
 
22
  Check: Final file state: the re-flattened message must equal msg_45.txt exactly, and the signed part must be byte-identical. No screenshot
23
  Tools: Python email package (message_from_file, Generator with maxheaderlen=0), or any MIME library that preserves the signed part; a script writes the re-flattened message to a file
24
  - `Lib/test/test_email/data/msg_21.txt`: Compose a two-part MIME message with a preamble and an epilogue from a spec.
25
  Format: RFC 5322 message (.txt)
26
  Origin: CPython email test data (tool_fixture)
 
 
27
  Check: Final file state: the composed message must equal msg_21.txt exactly (assertEqual on as_string). No screenshot
28
  Tools: Python email.mime (MIMEMultipart, MIMEText) and as_string, or mutt -H with a prepared draft; a script composes the message and writes it to a file
campaign_files/2026-10-gt-browser-tests-2/crisp-email-reply-parser-fixtures/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Each file is self-contained. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_values_in_test_code. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the German Outlook separator -----Ursprüngliche Nachricht-----; the script writes the first fragment to a file; email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the Danish separator -------- Oprindelig besked --------; the script writes the fragments to files
14
 
15
  **Licence (SPDX):** MIT
@@ -21,8 +19,6 @@ Files:
21
  - `test/fixtures/email_german_outlook.txt`: Keep only the new text above a German Outlook quote header.
22
  Format: plain-text message body (.txt); expected string in JavaScript test code
23
  Origin: crisp email-reply-parser test fixture (tool_fixture)
24
- Reference result: expected_values_in_test_code
25
- Reference source: test/test.js lines 290-298 (test_email_german_outlook): the first fragment, trimmed, equals COMMON_FIRST_FRAGMENT defined at lines 9-13
26
  Check: Final file state: the agent writes the reply fragment to a file; its content trimmed must equal the five-line lorem ipsum COMMON_FIRST_FRAGMENT of test/test.js lines 9-13 exactly. No screenshot
27
  Tools: email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the German Outlook separator -----Ursprüngliche Nachricht-----; the script writes the first fragment to a file
28
  Companions (not included): test/test.js
@@ -30,8 +26,6 @@ Files:
30
  - `test/fixtures/email_danish_dash_separator.txt`: Split a reply from a Danish 'Oprindelig besked' quote block.
31
  Format: plain-text message body (.txt); expected values in JavaScript test code
32
  Origin: crisp email-reply-parser test fixture (tool_fixture)
33
- Reference result: expected_values_in_test_code
34
- Reference source: test/test.js lines 578-587 (test_email_original_message_danish_dash): two fragments, the first trimmed equals COMMON_FIRST_FRAGMENT (lines 9-13)
35
  Check: Final file state: the agent writes each fragment to its own file; there must be exactly 2 fragment files and the first, trimmed, must equal COMMON_FIRST_FRAGMENT of test/test.js lines 9-13. No screenshot
36
  Tools: email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the Danish separator -------- Oprindelig besked --------; the script writes the fragments to files
37
  Companions (not included): test/test.js
 
8
 
9
  **Self-contained:** Each file is self-contained. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the German Outlook separator -----Ursprüngliche Nachricht-----; the script writes the first fragment to a file; email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the Danish separator -------- Oprindelig besked --------; the script writes the fragments to files
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `test/fixtures/email_german_outlook.txt`: Keep only the new text above a German Outlook quote header.
20
  Format: plain-text message body (.txt); expected string in JavaScript test code
21
  Origin: crisp email-reply-parser test fixture (tool_fixture)
 
 
22
  Check: Final file state: the agent writes the reply fragment to a file; its content trimmed must equal the five-line lorem ipsum COMMON_FIRST_FRAGMENT of test/test.js lines 9-13 exactly. No screenshot
23
  Tools: email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the German Outlook separator -----Ursprüngliche Nachricht-----; the script writes the first fragment to a file
24
  Companions (not included): test/test.js
 
26
  - `test/fixtures/email_danish_dash_separator.txt`: Split a reply from a Danish 'Oprindelig besked' quote block.
27
  Format: plain-text message body (.txt); expected values in JavaScript test code
28
  Origin: crisp email-reply-parser test fixture (tool_fixture)
 
 
29
  Check: Final file state: the agent writes each fragment to its own file; there must be exactly 2 fragment files and the first, trimmed, must equal COMMON_FIRST_FRAGMENT of test/test.js lines 9-13. No screenshot
30
  Tools: email-reply-parser of crisp-oss (Node.js) getFragments, or a regex for the Danish separator -------- Oprindelig besked --------; the script writes the fragments to files
31
  Companions (not included): test/test.js
campaign_files/2026-10-gt-browser-tests-2/data-ical-parse/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_values_in_test_code. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Python icalendar or Perl Data::ICal: a script parses the file and writes calendar name, product id, event summary (unescaped), location, URL, alarm count and the time zone's DAYLIGHT and STANDARD counts to a JSON file
14
 
15
  **Licence (SPDX):** Artistic-1.0-Perl OR GPL-1.0-or-later
@@ -21,8 +19,6 @@ Files:
21
  - `t/ics/test.ics`: Read event fields, alarm and time zone rules from an iCalendar file with escaped text.
22
  Format: iCalendar (.ics); expected values in a Perl Test::More file
23
  Origin: Data::ICal test fixture (tool_fixture)
24
- Reference result: expected_values_in_test_code
25
- Reference source: t/08.parse.t lines 56-97: VCALENDAR with prodid 'Data::ICal test', calendar name 'Data::ICal test calendar', 2 entries; event summary 'Data::ICal release party,other things with slash\es' after unescaping, location 'The Restaurant at the End of the Universe', URL http://www.bestpractical.com with VALUE=URI, 1 audio alarm; time zone Europe/London with 6 sub-entries (3 DAYLIGHT, 3 STANDARD)
26
  Check: Final file state: each field of the written JSON must equal the expected value exactly; the summary must be unescaped (comma and backslash restored). No screenshot
27
  Tools: Python icalendar or Perl Data::ICal: a script parses the file and writes calendar name, product id, event summary (unescaped), location, URL, alarm count and the time zone's DAYLIGHT and STANDARD counts to a JSON file
28
  Companions (not included): t/08.parse.t; README
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Python icalendar or Perl Data::ICal: a script parses the file and writes calendar name, product id, event summary (unescaped), location, URL, alarm count and the time zone's DAYLIGHT and STANDARD counts to a JSON file
12
 
13
  **Licence (SPDX):** Artistic-1.0-Perl OR GPL-1.0-or-later
 
19
  - `t/ics/test.ics`: Read event fields, alarm and time zone rules from an iCalendar file with escaped text.
20
  Format: iCalendar (.ics); expected values in a Perl Test::More file
21
  Origin: Data::ICal test fixture (tool_fixture)
 
 
22
  Check: Final file state: each field of the written JSON must equal the expected value exactly; the summary must be unescaped (comma and backslash restored). No screenshot
23
  Tools: Python icalendar or Perl Data::ICal: a script parses the file and writes calendar name, product id, event summary (unescaped), location, URL, alarm count and the time zone's DAYLIGHT and STANDARD counts to a JSON file
24
  Companions (not included): t/08.parse.t; README
campaign_files/2026-10-gt-browser-tests-2/data-ical-vcal10/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_values_in_test_code. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Python vobject or Perl Data::ICal with vcal10 => 1: a script parses the 1.0 file and writes product id, entry count, summary and the list of properties it does not recognise to a JSON file
14
 
15
  **Licence (SPDX):** Artistic-1.0-Perl OR GPL-1.0-or-later
@@ -21,8 +19,6 @@ Files:
21
  - `t/ics/version1.ics`: Read an old vCalendar 1.0 export and flag the properties a 2.0 parser does not know.
22
  Format: vCalendar 1.0 file (.ics); expected values in a Perl Test::More file
23
  Origin: Data::ICal test fixture (tool_fixture)
24
- Reference result: expected_values_in_test_code
25
- Reference source: t/12.verson1.t lines 13-82: without the vCalendar 1.0 switch the file is rejected; with it the calendar has prodid '-//Mirapoint Calendar', 1 entry, summary 'cal1' and 0 sub-entries; DCREATED and MALARM are reported as unknown properties
26
  Check: Final file state: the written JSON must give prodid "-//Mirapoint Calendar", 1 entry, summary "cal1", 0 sub-entries and the unknown properties {dcreated, malarm}. No screenshot
27
  Tools: Python vobject or Perl Data::ICal with vcal10 => 1: a script parses the 1.0 file and writes product id, entry count, summary and the list of properties it does not recognise to a JSON file
28
  Companions (not included): t/12.verson1.t; README
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Python vobject or Perl Data::ICal with vcal10 => 1: a script parses the 1.0 file and writes product id, entry count, summary and the list of properties it does not recognise to a JSON file
12
 
13
  **Licence (SPDX):** Artistic-1.0-Perl OR GPL-1.0-or-later
 
19
  - `t/ics/version1.ics`: Read an old vCalendar 1.0 export and flag the properties a 2.0 parser does not know.
20
  Format: vCalendar 1.0 file (.ics); expected values in a Perl Test::More file
21
  Origin: Data::ICal test fixture (tool_fixture)
 
 
22
  Check: Final file state: the written JSON must give prodid "-//Mirapoint Calendar", 1 entry, summary "cal1", 0 sub-entries and the unknown properties {dcreated, malarm}. No screenshot
23
  Tools: Python vobject or Perl Data::ICal with vcal10 => 1: a script parses the 1.0 file and writes product id, entry count, summary and the list of properties it does not recognise to a JSON file
24
  Companions (not included): t/12.verson1.t; README
campaign_files/2026-10-gt-browser-tests-2/deltachat-provider-db/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_account_server_settings. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
14
 
15
  **Licence (SPDX):** MPL-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `_providers/disroot.md`: Mail account setup for a Disroot address: IMAP SSL 993 and SMTP STARTTLS 587 with the local part as user name.
22
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
23
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
24
- Reference result: expected_account_server_settings
25
- Reference source: _providers/disroot.md lines 1-19 (front matter): status OK; last_checked 2017-06; domain disroot.org; imap socket SSL hostname disroot.org port 993 username_pattern EMAILLOCALPART; smtp socket STARTTLS hostname disroot.org port 587 username_pattern EMAILLOCALPART; field meanings and the username_pattern default (EMAIL) in README.md lines 36-70
26
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); user name is the local part (EMAILLOCALPART), not the full address
27
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
28
  Companions (not included): README.md; LICENSE
@@ -30,8 +26,6 @@ Files:
30
  - `_providers/fastmail.md`: Mail account setup for a Fastmail address: IMAP SSL 993 and SMTP SSL 465, app password required.
31
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
32
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
33
- Reference result: expected_account_server_settings
34
- Reference source: _providers/fastmail.md lines 1-140 (front matter): status PREPARATION; 117 domains listed (lines 6-122); imap socket SSL hostname imap.fastmail.com port 993; smtp socket SSL hostname smtp.fastmail.com port 465 (lines 125-132; the STARTTLS 587 block at lines 133-137 is commented out as very strongly discouraged); no username_pattern, so the README default EMAIL applies; before_login_hint (line 138) says an app-specific password must be created first; the markdown body (lines 142-152) gives the web-interface steps; field meanings and the username_pattern default (EMAIL) in README.md lines 36-70
35
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); commented-out STARTTLS block is not an accepted answer
36
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
37
  Companions (not included): README.md; LICENSE
@@ -39,8 +33,6 @@ Files:
39
  - `_providers/gmx.net.md`: Mail account setup for a GMX address: IMAP SSL 993 and SMTP SSL 465 or STARTTLS 587, IMAP access to be enabled first.
40
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
41
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
42
- Reference result: expected_account_server_settings
43
- Reference source: _providers/gmx.net.md lines 1-31 (front matter): status PREPARATION; domains gmx.net, gmx.de, gmx.at, gmx.ch, gmx.org, gmx.eu, gmx.info, gmx.biz, gmx.com; imap SSL imap.gmx.net 993; smtp SSL mail.gmx.net 465 or STARTTLS mail.gmx.net 587; no username_pattern, so the README default EMAIL applies; before_login_hint: IMAP access must be allowed first; last_checked 2017-06; field meanings and the username_pattern default (EMAIL) in README.md lines 36-70
44
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); either SMTP alternative is accepted
45
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
46
  Companions (not included): README.md; LICENSE
@@ -48,8 +40,6 @@ Files:
48
  - `_providers/migadu.md`: Mail account setup for a Migadu address: IMAP SSL 993 and SMTP SSL 465 or STARTTLS 587 with the full address as user name.
49
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
50
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
51
- Reference result: expected_account_server_settings
52
- Reference source: _providers/migadu.md lines 1-24 (front matter): status OK; domain migadu.com; imap SSL imap.migadu.com 993; smtp SSL smtp.migadu.com 465 or STARTTLS smtp.migadu.com 587; username_pattern EMAIL on every server; last_checked 2024-09; field meanings and the username_pattern default (EMAIL) in README.md lines 36-70
53
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); either SMTP alternative is accepted
54
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
55
  Companions (not included): README.md; LICENSE
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
12
 
13
  **Licence (SPDX):** MPL-2.0
 
19
  - `_providers/disroot.md`: Mail account setup for a Disroot address: IMAP SSL 993 and SMTP STARTTLS 587 with the local part as user name.
20
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
21
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
 
 
22
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); user name is the local part (EMAILLOCALPART), not the full address
23
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
24
  Companions (not included): README.md; LICENSE
 
26
  - `_providers/fastmail.md`: Mail account setup for a Fastmail address: IMAP SSL 993 and SMTP SSL 465, app password required.
27
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
28
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
 
 
29
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); commented-out STARTTLS block is not an accepted answer
30
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
31
  Companions (not included): README.md; LICENSE
 
33
  - `_providers/gmx.net.md`: Mail account setup for a GMX address: IMAP SSL 993 and SMTP SSL 465 or STARTTLS 587, IMAP access to be enabled first.
34
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
35
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
 
 
36
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); either SMTP alternative is accepted
37
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
38
  Companions (not included): README.md; LICENSE
 
40
  - `_providers/migadu.md`: Mail account setup for a Migadu address: IMAP SSL 993 and SMTP SSL 465 or STARTTLS 587 with the full address as user name.
41
  Format: Markdown file with YAML front matter (name, status, domains, server list with type, socket, hostname, port, username_pattern; optional hints and a markdown help page)
42
  Origin: real mail provider settings curated by the Delta Chat project for account setup; README lines 3-10 say the database is archived and reflects the state of May 2026
 
 
43
  Check: Final configuration state, no screenshot needed: read the agent-created account from the Thunderbird profile prefs.js (mail.server.serverN.hostname, port, socketType, userName, type and the mail.smtpserver.smtpN.* lines) and check that hostname and port equal the provider file exactly, that the connection security equals the socket value (SSL maps to socketType 3, STARTTLS to 2), and that userName follows username_pattern (EMAIL is the full address, EMAILLOCALPART the part before @); either SMTP alternative is accepted
44
  Tools: write the account into the Thunderbird profile's prefs.js (mail.server.serverN.* and mail.smtpserver.smtpN.* user_pref lines) with a text editor or script while Thunderbird is closed, or into another client's text config (for example aerc accounts.conf or a mutt or neomutt muttrc); the provider file itself is YAML front matter read with any YAML parser
45
  Companions (not included): README.md; LICENSE
campaign_files/2026-10-gt-browser-tests-2/docx2python-test-resources/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: filled_grid_text; checkbox_state. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** docx2python (duplicate_merged_cells=True) or python-docx or an XML parser over w:gridSpan and w:vMerge; the agent writes grid.csv with the merged text repeated in every covered position (and grid_blank.csv with only the origin filled); docx2python, or an XML parser over w:ffData/w:checkBox (w:checked present without w:val, or w:val 1 or true, means checked; w:default as fallback); the agent writes checked.json with the labels of the checked boxes in document order and the counts of checked and unchecked boxes
14
 
15
  **Licence (SPDX):** MIT
@@ -21,8 +19,6 @@ Files:
21
  - `tests/resources/merged_cells.docx`: Expand a Word table with horizontal and vertical merges into a 5 by 4 grid.
22
  Format: Word document (.docx)
23
  Origin: docx2python test resource authored in Word (tool_fixture)
24
- Reference result: filled_grid_text
25
- Reference source: tests/test_merged_cells.py lines 34-43 (duplicated grid: rows 0-0 0-12 0-12 0-3 / 12-0 1-1 1-2 1-3 / 12-0 2-1 2-2 2-3 / 3-0 34-123 34-123 34-123 / 4-0 34-123 34-123 34-123) and lines 18-27 (origin-only grid with blanks)
26
  Check: Final file state: read grid.csv (and grid_blank.csv) and compare cell by cell with the two test grids. No screenshot.
27
  Tools: docx2python (duplicate_merged_cells=True) or python-docx or an XML parser over w:gridSpan and w:vMerge; the agent writes grid.csv with the merged text repeated in every covered position (and grid_blank.csv with only the origin filled)
28
  Companions (not included): tests/test_merged_cells.py (the upstream test; both expected grids live in its asserts, hide from the agent); tests/conftest.py (defines RESOURCES used by the test, hide from the agent)
@@ -30,8 +26,6 @@ Files:
30
  - `tests/resources/checked_boxes.docx`: List the checked and count the unchecked form checkboxes of a blank Word referral form.
31
  Format: Word document (.docx, legacy form fields)
32
  Origin: user-submitted blank referral form (an elder abuse forensic center referral template) kept as a docx2python test resource (tool_fixture); every form text field is empty
33
- Reference result: checkbox_state
34
- Reference source: tests/test_checked_boxes.py lines 49-93 (the 12 checked labels in section 1: Adult Protective Services ... Other (describe)) and lines 123-124 (12 checked, 32 unchecked)
35
  Check: Final file state: read checked.json; compare the checked label list (stripped of the leading space and trailing en spaces) and the two counts with the test. No screenshot.
36
  Tools: docx2python, or an XML parser over w:ffData/w:checkBox (w:checked present without w:val, or w:val 1 or true, means checked; w:default as fallback); the agent writes checked.json with the labels of the checked boxes in document order and the counts of checked and unchecked boxes
37
  Companions (not included): tests/test_checked_boxes.py (the upstream test; the checked labels and the counts live in its asserts, hide from the agent); tests/conftest.py (defines RESOURCES used by the test, hide from the agent)
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 2 of 2 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** docx2python (duplicate_merged_cells=True) or python-docx or an XML parser over w:gridSpan and w:vMerge; the agent writes grid.csv with the merged text repeated in every covered position (and grid_blank.csv with only the origin filled); docx2python, or an XML parser over w:ffData/w:checkBox (w:checked present without w:val, or w:val 1 or true, means checked; w:default as fallback); the agent writes checked.json with the labels of the checked boxes in document order and the counts of checked and unchecked boxes
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `tests/resources/merged_cells.docx`: Expand a Word table with horizontal and vertical merges into a 5 by 4 grid.
20
  Format: Word document (.docx)
21
  Origin: docx2python test resource authored in Word (tool_fixture)
 
 
22
  Check: Final file state: read grid.csv (and grid_blank.csv) and compare cell by cell with the two test grids. No screenshot.
23
  Tools: docx2python (duplicate_merged_cells=True) or python-docx or an XML parser over w:gridSpan and w:vMerge; the agent writes grid.csv with the merged text repeated in every covered position (and grid_blank.csv with only the origin filled)
24
  Companions (not included): tests/test_merged_cells.py (the upstream test; both expected grids live in its asserts, hide from the agent); tests/conftest.py (defines RESOURCES used by the test, hide from the agent)
 
26
  - `tests/resources/checked_boxes.docx`: List the checked and count the unchecked form checkboxes of a blank Word referral form.
27
  Format: Word document (.docx, legacy form fields)
28
  Origin: user-submitted blank referral form (an elder abuse forensic center referral template) kept as a docx2python test resource (tool_fixture); every form text field is empty
 
 
29
  Check: Final file state: read checked.json; compare the checked label list (stripped of the leading space and trailing en spaces) and the two counts with the test. No screenshot.
30
  Tools: docx2python, or an XML parser over w:ffData/w:checkBox (w:checked present without w:val, or w:val 1 or true, means checked; w:default as fallback); the agent writes checked.json with the labels of the checked boxes in document order and the counts of checked and unchecked boxes
31
  Companions (not included): tests/test_checked_boxes.py (the upstream test; the checked labels and the counts live in its asserts, hide from the agent); tests/conftest.py (defines RESOURCES used by the test, hide from the agent)
campaign_files/2026-10-gt-browser-tests-2/dovecot-pigeonhole-tests/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_filter_actions. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** write the filter as a .sieve file with a text editor and upload it with a ManageSieve client (sieve-connect) or place it as the user's active script; check it offline with Pigeonhole's sieve-test command (sieve-test script.sieve message.eml prints the actions it would take) or run the upstream testsuite tool on actions.svtest
14
 
15
  **Licence (SPDX):** LGPL-2.1-only
@@ -21,8 +19,6 @@ Files:
21
  - `tests/execute/actions.svtest`: Mail filter rules in Sieve: file into folders and redirect, with the expected list of resulting actions.
22
  Format: Sieve testsuite file (vnd.dovecot.testsuite) plus two RFC 5228 Sieve filter scripts
23
  Origin: Pigeonhole testsuite file and its two filter scripts (tool_fixture)
24
- Reference result: expected_filter_actions
25
- Reference source: tests/execute/actions.svtest: the test message (lines 5-12, To nico@frop.example.org), test "Fileinto" (lines 17-45): fileinto.sieve must yield exactly 3 actions, store, store, keep, in that order; test "Redirect" (lines 47-79): redirect.sieve must yield exactly 4 actions, redirect, keep, redirect, redirect; the filter rules are in tests/execute/actions/fileinto.sieve and redirect.sieve (duplicates of earlier actions are merged)
26
  Check: Final configuration state, no screenshot needed: take the agent's Sieve script, run it on the test message with sieve-test or the upstream testsuite, and compare the action list with the svtest assertions (exact action count and action kind at each index); for the folder names and redirect targets, compare with the reference scripts (INBOX.VB, INBOX.backup and keep; stephan@example.com, keep, stephan@example.org, nico@example.nl)
27
  Tools: write the filter as a .sieve file with a text editor and upload it with a ManageSieve client (sieve-connect) or place it as the user's active script; check it offline with Pigeonhole's sieve-test command (sieve-test script.sieve message.eml prints the actions it would take) or run the upstream testsuite tool on actions.svtest
28
  Companions (not included): tests/execute/actions/fileinto.sieve; tests/execute/actions/redirect.sieve; COPYING; COPYING.LGPL
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** write the filter as a .sieve file with a text editor and upload it with a ManageSieve client (sieve-connect) or place it as the user's active script; check it offline with Pigeonhole's sieve-test command (sieve-test script.sieve message.eml prints the actions it would take) or run the upstream testsuite tool on actions.svtest
12
 
13
  **Licence (SPDX):** LGPL-2.1-only
 
19
  - `tests/execute/actions.svtest`: Mail filter rules in Sieve: file into folders and redirect, with the expected list of resulting actions.
20
  Format: Sieve testsuite file (vnd.dovecot.testsuite) plus two RFC 5228 Sieve filter scripts
21
  Origin: Pigeonhole testsuite file and its two filter scripts (tool_fixture)
 
 
22
  Check: Final configuration state, no screenshot needed: take the agent's Sieve script, run it on the test message with sieve-test or the upstream testsuite, and compare the action list with the svtest assertions (exact action count and action kind at each index); for the folder names and redirect targets, compare with the reference scripts (INBOX.VB, INBOX.backup and keep; stephan@example.com, keep, stephan@example.org, nico@example.nl)
23
  Tools: write the filter as a .sieve file with a text editor and upload it with a ManageSieve client (sieve-connect) or place it as the user's active script; check it offline with Pigeonhole's sieve-test command (sieve-test script.sieve message.eml prints the actions it would take) or run the upstream testsuite tool on actions.svtest
24
  Companions (not included): tests/execute/actions/fileinto.sieve; tests/execute/actions/redirect.sieve; COPYING; COPYING.LGPL
campaign_files/2026-10-hw-eda-tests-2/cq-kit-at-commit-2b3897b-github-michaelgale-cq-kit-a-cadquery-helper-0061fa24/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: cad_geometry_assert. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_basic.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI; python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_helpers.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI; python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_xsection.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI; python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_ribbon.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
14
 
15
  **Licence (SPDX):** MIT
@@ -21,8 +19,6 @@ Files:
21
  - `tests/test_basic.py`: Cq-kit basic shape tests: drafted boxes and cylinders, rounded and recentred solids; check the computed 3D and 2D sizes.
22
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
23
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
24
- Reference result: cad_geometry_assert
25
- Reference source: upstream expected values written in the test file itself: line 28 `assert _almost_same(size_3d(r), (1, 2, 3))`; line 30 `assert _almost_same(size_3d(r), (2.440, 4.440, 5))` (box drafted by an angle); line 31 face size (2.441, 4.441).
26
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
27
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_basic.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
28
  Companions (not included): LICENSE
@@ -30,8 +26,6 @@ Files:
30
  - `tests/test_helpers.py`: Cq-kit helper tests: rotate, extrude, recentre, realign, multi-section extrude, inverse fillet and chamfer; check the resulting bounding boxes.
31
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
32
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
33
- Reference result: cad_geometry_assert
34
- Reference source: upstream expected values written in the test file itself: line 17 `assert _almost_same((mx, my, mz, px, py, pz), (-1.5, -4, -2.5, 1.5, 0, 2.5))` (bounding box after rotation); 86 _almost_same checks on bounds and sizes.
35
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
36
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_helpers.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
37
  Companions (not included): LICENSE
@@ -39,8 +33,6 @@ Files:
39
  - `tests/test_xsection.py`: Cq-kit cross-section tests: aluminium extrusion profiles (2020, 3030, 4040, 2040, 3060) rendered and extruded; check profile points and extruded sizes.
40
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
41
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
42
- Reference result: cad_geometry_assert
43
- Reference source: upstream expected values written in the test file itself: lines 183-186 `xc = XSection.xs2020("XY")` ... `assert _almost_same(bb, (20, 20, 400))`; lines 203-206 xs2040 on ZY extruded 200 gives (200, 40, 20); profile point checks from line 97.
44
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
45
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_xsection.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
46
  Companions (not included): LICENSE
@@ -48,8 +40,6 @@ Files:
48
  - `tests/test_ribbon.py`: Cq-kit ribbon tests: a ribbon outline drawn from turtle-style path commands; check the wire length and the face and edge counts after extrusion.
49
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
50
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
51
- Reference result: cad_geometry_assert
52
- Reference source: upstream expected values written in the test file itself: line 56 `assert abs(wire_length(wire) - 24.622) < 1e-3`; lines 58-62 extruded by 1 gives 14 faces and 36 edges; line 69 same length for the second path.
53
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
54
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_ribbon.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
55
  Companions (not included): LICENSE
 
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 4 of 4 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_basic.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI; python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_helpers.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI; python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_xsection.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI; python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_ribbon.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `tests/test_basic.py`: Cq-kit basic shape tests: drafted boxes and cylinders, rounded and recentred solids; check the computed 3D and 2D sizes.
20
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
21
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
 
 
22
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
23
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_basic.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
24
  Companions (not included): LICENSE
 
26
  - `tests/test_helpers.py`: Cq-kit helper tests: rotate, extrude, recentre, realign, multi-section extrude, inverse fillet and chamfer; check the resulting bounding boxes.
27
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
28
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
 
 
29
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
30
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_helpers.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
31
  Companions (not included): LICENSE
 
33
  - `tests/test_xsection.py`: Cq-kit cross-section tests: aluminium extrusion profiles (2020, 3030, 4040, 2040, 3060) rendered and extruded; check profile points and extruded sizes.
34
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
35
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
 
 
36
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
37
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_xsection.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
38
  Companions (not included): LICENSE
 
40
  - `tests/test_ribbon.py`: Cq-kit ribbon tests: a ribbon outline drawn from turtle-style path commands; check the wire length and the face and edge counts after extrusion.
41
  Format: Python pytest module on CadQuery and cq-kit (OpenCASCADE kernel)
42
  Origin: library self-test written by the library's developers; expected numbers are closed-form geometry (sizes, diameters, volumes) or recorded kernel results
 
 
43
  Check: Task: given the modelling steps of each test function (primitive sizes, sketch or profile points, extrude lengths, draft angles) with the expected numbers hidden, build the shape and report the measured sizes, bounds or lengths, or derive them in closed form. Judge: compare each reported number with the test's expected value by the same rule as the upstream assertion. Answer is the test file's own expected values.
44
  Tools: python3 with CadQuery and cq-kit installed runs the module under pytest headless (pytest tests/test_ribbon.py); sizes read with cqkit size_3d, bounds and wire_length on OpenCASCADE shapes; no GUI
45
  Companions (not included): LICENSE
campaign_files/2026-10-m-issue-trackers-2/coreboot-coreboot/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m209-12/thread_NBBLCADHUGKYSZXD7YLD6JVHJWOVNS5Z_emails.json`: [coreboot] BIOS chip address range? (8 messages, 2019-03-27 to 2019-03-29).
22
  Format: HyperKitty REST API JSON (?format=json): list of the thread's messages, each with message_id, message_id_hash, subject, sender (address obfuscated with '(a)', mailman_id), sender_name, date (with zone), parent (URL of the parent message, null for the root), children and thread.
23
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one thread (all its messages, across months), saved as the JSON the API returns; the listing is the pinned pointer.
24
- Reference result: facts computed from the thread headers
25
- Reference source: message_id, parent, sender (mailman_id), sender_name and date fields of the thread listing, read from the pinned JSON.
26
  Check: Exact match against values computed from the pinned JSON: count the entries; sort by date converted to UTC; a direct reply is an entry whose parent URL ends with the root's message_id_hash; the longest chain and branch points follow the parent links (cross-checked against the children lists, which agree). People are counted by sender.mailman_id, falling back to the obfuscated sender address when mailman_id is null (messages imported from the old archive); on this thread that count equals the count by sender_name after removing accents and letter case. The first-reply interval is the exact difference in whole seconds between the dates of the root and the earliest reply. Deterministic. Hand-checked on 2 threads of this host (general rule 5): the chosen thread and '[coreboot] OpenBSD won't install on an x220 Thinkpad with Coreboot' (thread EV2GK4ZHQS6KWM4BFSD4CV6JC5VZNNCM, 5 messages); counts, last sender and time, first-reply seconds, direct replies and parent-against-children links read by hand agree with the computed facts.
27
  Tools: Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m209-12/thread_NBBLCADHUGKYSZXD7YLD6JVHJWOVNS5Z_emails.json`: [coreboot] BIOS chip address range? (8 messages, 2019-03-27 to 2019-03-29).
20
  Format: HyperKitty REST API JSON (?format=json): list of the thread's messages, each with message_id, message_id_hash, subject, sender (address obfuscated with '(a)', mailman_id), sender_name, date (with zone), parent (URL of the parent message, null for the root), children and thread.
21
  Origin: Real list traffic. The unit is the HyperKitty REST listing of one thread (all its messages, across months), saved as the JSON the API returns; the listing is the pinned pointer.
 
 
22
  Check: Exact match against values computed from the pinned JSON: count the entries; sort by date converted to UTC; a direct reply is an entry whose parent URL ends with the root's message_id_hash; the longest chain and branch points follow the parent links (cross-checked against the children lists, which agree). People are counted by sender.mailman_id, falling back to the obfuscated sender address when mailman_id is null (messages imported from the old archive); on this thread that count equals the count by sender_name after removing accents and letter case. The first-reply interval is the exact difference in whole seconds between the dates of the root and the earliest reply. Deterministic. Hand-checked on 2 threads of this host (general rule 5): the chosen thread and '[coreboot] OpenBSD won't install on an x220 Thinkpad with Coreboot' (thread EV2GK4ZHQS6KWM4BFSD4CV6JC5VZNNCM, 5 messages); counts, last sender and time, first-reply seconds, direct replies and parent-against-children links read by hand agree with the computed facts.
23
  Tools: Fetch or read the JSON with Python's json module (or jq); rebuild the reply tree from the parent links. The same thread can be exported as mbox from HyperKitty and read with mutt -f.
campaign_files/2026-10-m-issue-trackers-2/coreboot-gerrit-changes/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.coreboot.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `changes/86699/detail.json`: Coreboot change 86699: mb/google/fatcat/var/francka: Add support for DMIC0.
22
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
23
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments)
25
- Reference source: The pinned response: changes/86699/detail.json.
26
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: hardwaretestrobot, jenkins, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Gerrit REST API (GET https://review.coreboot.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.coreboot.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `changes/86699/detail.json`: Coreboot change 86699: mb/google/fatcat/var/francka: Add support for DMIC0.
20
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
21
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: hardwaretestrobot, jenkins, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Gerrit REST API (GET https://review.coreboot.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
campaign_files/2026-10-m-issue-trackers-2/couchbase-gerrit-changes/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.couchbase.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `changes/228257/detail.json`: Couchbase-net-client change 228257: NCBC-3738 Support Binary Objects in Transactions.
22
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
23
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments)
25
- Reference source: The pinned response: changes/228257/detail.json.
26
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: buildbot, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Gerrit REST API (GET https://review.couchbase.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Gerrit REST API (GET https://review.couchbase.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `changes/228257/detail.json`: Couchbase-net-client change 228257: NCBC-3738 Support Binary Objects in Transactions.
20
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
21
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: buildbot, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Gerrit REST API (GET https://review.couchbase.org/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
campaign_files/2026-10-m-issue-trackers-2/courtney-passenger-gtfs/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the service calendar itself (active services on a date, removed or added dates, day counts). Where the reference result comes from and how a solution is checked is written per unit in the notes.
12
-
13
  **Tools to open and check:** Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
14
 
15
  **Licence (SPDX):** OGL-UK-3.0
@@ -22,8 +20,6 @@ Units:
22
  Read out of the archive: https://s3-eu-west-1.amazonaws.com/passenger-sources/courtney/gtfs/courtney_1785422871.zip
23
  Format: text/csv (GTFS Schedule calendar.txt; companions calendar_dates.txt, agency.txt)
24
  Origin: Live static GTFS feed published by Thames Valley Buses (published through Passenger, data.discoverpassenger.com) at https://s3-eu-west-1.amazonaws.com/passenger-sources/courtney/gtfs/courtney_1785422871.zip. calendar.txt and its companions were read by member out of the zip with HTTP Range requests (spec general rule 4: by member); kept as a snapshot pinned by sha256.
25
- Reference result: facts computed from the service calendar itself (active services on a date, removed or added dates, day counts)
26
- Reference source: calendar.txt weekday flags with start_date and end_date, calendar_dates.txt exception rows, agency_timezone of agency.txt, all from the pinned members.
27
  Check: deterministic, no model judgement. Hand-checking 2 samples per source is pending for the delivery step (spec general rule 5).
28
  Tools: Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
29
  Companions (not included): calendar_dates.txt; agency.txt
 
8
 
9
  **Self-contained:** The unit is self-contained. 1 of 1 units name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per unit in the notes.
10
 
 
 
11
  **Tools to open and check:** Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
12
 
13
  **Licence (SPDX):** OGL-UK-3.0
 
20
  Read out of the archive: https://s3-eu-west-1.amazonaws.com/passenger-sources/courtney/gtfs/courtney_1785422871.zip
21
  Format: text/csv (GTFS Schedule calendar.txt; companions calendar_dates.txt, agency.txt)
22
  Origin: Live static GTFS feed published by Thames Valley Buses (published through Passenger, data.discoverpassenger.com) at https://s3-eu-west-1.amazonaws.com/passenger-sources/courtney/gtfs/courtney_1785422871.zip. calendar.txt and its companions were read by member out of the zip with HTTP Range requests (spec general rule 4: by member); kept as a snapshot pinned by sha256.
 
 
23
  Check: deterministic, no model judgement. Hand-checking 2 samples per source is pending for the delivery step (spec general rule 5).
24
  Tools: Read the CSV files with Python csv or pandas (or sqlite3 .import) and expand service days per GTFS rules; gtfs-kit or partridge compute active service_ids per date headless; dates are local to agency_timezone (Python zoneinfo).
25
  Companions (not included): calendar_dates.txt; agency.txt
campaign_files/2026-10-m-issue-trackers-2/dart-gerrit-changes/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Gerrit REST API (GET https://dart-review.googlesource.com/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `changes/431703/detail.json`: Sdk change 431703: analyzer: rename Lint*Context classes.
22
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
23
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments)
25
- Reference source: The pinned response: changes/431703/detail.json.
26
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: dart-scoped@luci-project-accounts.iam.gserviceaccount.com, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Gerrit REST API (GET https://dart-review.googlesource.com/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Gerrit REST API (GET https://dart-review.googlesource.com/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `changes/431703/detail.json`: Sdk change 431703: analyzer: rename Lint*Context classes.
20
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote (DETAILED_LABELS), messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
21
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: dart-scoped@luci-project-accounts.iam.gserviceaccount.com, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Gerrit REST API (GET https://dart-review.googlesource.com/changes/{number}/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
campaign_files/2026-10-m-issue-trackers-2/dawn-gerrit-changes/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Gerrit REST API (GET https://dawn-review.googlesource.com/changes/298275/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `changes/298275/detail.json`: Dawn change 298275: Reland "[dawn] Switch indirect dispatch validation to use immediates".
22
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote, messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
23
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the change's own review timeline (status, submit time and submitter, patch sets, votes, reviewers added or removed, review comments)
25
- Reference source: The pinned response: changes/298275/detail.json.
26
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: dawn-scoped@luci-project-accounts.iam.gserviceaccount.com, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler|maloo|janitor|autotest|gerrit). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Gerrit REST API (GET https://dawn-review.googlesource.com/changes/298275/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Gerrit REST API (GET https://dawn-review.googlesource.com/changes/298275/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `changes/298275/detail.json`: Dawn change 298275: Reland "[dawn] Switch indirect dispatch validation to use immediates".
20
  Format: Gerrit REST API ChangeInfo JSON from the /detail endpoint, preceded by the anti-XSSI line ")]}'": status, created, updated, submitted, submitter, owner, labels with every vote, messages (author, real_author, date, tag, _revision_number, message text) and reviewer_updates (updated, updated_by, reviewer, state REVIEWER/CC/REMOVED). Patch-set uploads are messages tagged autogenerated:gerrit:newPatchSet; votes appear in message text as 'Patch Set N: Label+V' and in labels.*.all. The endpoint returns all messages and reviewer updates of the change, so the record is complete.
21
  Origin: Live Gerrit REST interface of a closed (merged) code review change, collected as a NOASSERTION pointer row under spec general rule 4: the detail JSON was fetched twice apart and was byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: status, submitted and submitter; reviewer_updates entries (reviewer, state, updated_by, updated); messages tagged autogenerated:gerrit:newPatchSet (patch-set count, _revision_number); votes from labels.*.all (value, _account_id) and from message first lines 'Patch Set N: Label+V'. A human review comment is a message whose author is not a bot or service account, whose tag does not start with 'autogenerated:', and whose text has content beyond the 'Patch Set N: votes' first line or names inline comments '(N comments)'. Bot and service accounts excluded from the comment count: dawn-scoped@luci-project-accounts.iam.gserviceaccount.com, accounts tagged SERVICE_USER, and any account whose name, username or email matches the pattern (bot|zuul|jenkins|ci|luci|gerrit code review|service user|automation|buildbot|lint|sonar|tryjob|cq|autosubmit|copybara|tricium|gerritforge|verifier|checker|scheduler|maloo|janitor|autotest|gerrit). Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Gerrit REST API (GET https://dawn-review.googlesource.com/changes/298275/detail), strip the first line ")]}'" and read the JSON with jq or Python json; the git-review CLI and Gerrit's ssh 'gerrit query --format=JSON --all-approvals --comments' reach the same data where ssh access is open
campaign_files/2026-10-m-issue-trackers-2/digium-asterisk-users/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m141-15/2019-March.txt.gz`: [asterisk-users] AMI mulitple calls quickly (5 messages, 2019-03-12 to 2019-03-12).
22
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
23
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
24
- Reference result: facts computed from the thread headers
25
- Reference source: Message-ID, In-Reply-To, References, From and Date headers of the messages in the thread, read from the pinned month file.
26
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[asterisk-users] Paging systems?' (15 messages, root <46FCA36BC7E1D546AC28870158F9C25B03D4024CE3@ABRAHAM.highpoweredhelp.local>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m141-15/2019-March.txt.gz`: [asterisk-users] AMI mulitple calls quickly (5 messages, 2019-03-12 to 2019-03-12).
20
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
21
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
 
 
22
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[asterisk-users] Paging systems?' (15 messages, root <46FCA36BC7E1D546AC28870158F9C25B03D4024CE3@ABRAHAM.highpoweredhelp.local>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
campaign_files/2026-10-m-issue-trackers-2/django-trac/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the ticket change feed. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Trac ticket RSS (?format=rss) or CSV (?format=csv) and parse with Python xml.etree / csv; Trac's XML-RPC plugin is an alternative where enabled.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m027-02/ticket_30000.rss`: Django Trac #30000: QuerySet constructed with .union() should raise an exception on unsupported filter() attempt (closed, duplicate, 2019-07-17).
22
  Format: Trac ticket RSS 2.0: one item per change (author, pubDate, description holding field changes such as owner, status, resolution, plus the comment text).
23
  Origin: three fetches over about 18 minutes byte-identical.
24
- Reference result: facts computed from the ticket change feed
25
- Reference source: RSS items (dc:creator, pubDate, description).
26
  Check: Exact match against values parsed from the pinned RSS items. Deterministic. Extraction program to be hand-checked on 2 Trac tickets before answers are issued (general rule 5); pending.
27
  Tools: curl the Trac ticket RSS (?format=rss) or CSV (?format=csv) and parse with Python xml.etree / csv; Trac's XML-RPC plugin is an alternative where enabled.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Trac ticket RSS (?format=rss) or CSV (?format=csv) and parse with Python xml.etree / csv; Trac's XML-RPC plugin is an alternative where enabled.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m027-02/ticket_30000.rss`: Django Trac #30000: QuerySet constructed with .union() should raise an exception on unsupported filter() attempt (closed, duplicate, 2019-07-17).
20
  Format: Trac ticket RSS 2.0: one item per change (author, pubDate, description holding field changes such as owner, status, resolution, plus the comment text).
21
  Origin: three fetches over about 18 minutes byte-identical.
 
 
22
  Check: Exact match against values parsed from the pinned RSS items. Deterministic. Extraction program to be hand-checked on 2 Trac tickets before answers are issued (general rule 5); pending.
23
  Tools: curl the Trac ticket RSS (?format=rss) or CSV (?format=csv) and parse with Python xml.etree / csv; Trac's XML-RPC plugin is an alternative where enabled.
campaign_files/2026-10-m-issue-trackers-2/dns-oarc-dns-operations/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m141-08/2019-March.txt.gz`: [dns-operations] Custom DNS server as backend for authoritative DNS server (13 messages, 2019-03-14 to 2019-03-21).
22
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
23
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
24
- Reference result: facts computed from the thread headers
25
- Reference source: Message-ID, In-Reply-To, References, From and Date headers of the messages in the thread, read from the pinned month file.
26
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[dns-operations] Can Root DNS server modify the response?' (32 messages, root <ab1a827b192daf6cbc11df36c51c922a.squirrel@giyzk7o6dcunb2ry.onion>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m141-08/2019-March.txt.gz`: [dns-operations] Custom DNS server as backend for authoritative DNS server (13 messages, 2019-03-14 to 2019-03-21).
20
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
21
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
 
 
22
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[dns-operations] Can Root DNS server modify the response?' (32 messages, root <ab1a827b192daf6cbc11df36c51c922a.squirrel@giyzk7o6dcunb2ry.onion>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
campaign_files/2026-10-m-issue-trackers-2/dotnet-runtime-github-pulls/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the pull request's own timeline (status changes, review requests and assignments, comments and their authors). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl -H 'Accept: application/vnd.github+json' https://api.github.com/repos/dotnet/runtime/issues/49947/events?per_page=100 and .../issues/49947/comments?per_page=100, merge by created_at with jq or Python json; 'gh api' reaches the same endpoints
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,8 +19,6 @@ Files:
21
  - `repos/dotnet/runtime/issues/49947/events.json`: dotnet/runtime pull request #49947: Fix DEBUG/non-DEBUG asm diffs due to instruction group size estimation.
22
  Format: Two GitHub REST API JSON files fetched from the same issue: events.json (GET /repos/{owner}/{repo}/issues/{number}/events, per_page=100: actor, event, created_at, assignee/assigner, requested_reviewer/review_requester, label, commit_id; 25 entries) and the companion comments.json (GET /repos/{owner}/{repo}/issues/{number}/comments, per_page=100: user, created_at, updated_at, author_association, body; 12 entries). Each is one page with fewer than 100 entries, so both are complete. Together they hold the issue timeline without 'cross-referenced' entries, which embed live repository counters and are not byte-stable (see report). Pull request review submissions and inline review comments (pulls/{number}/reviews and pulls/{number}/comments) are not part of this unit.
23
  Origin: Live GitHub REST interface of a closed pull request, collected as a NOASSERTION pointer row under spec general rule 4: the events and comments JSON were each fetched twice apart and were byte-identical. Unit files kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the pull request's own timeline (status changes, review requests and assignments, comments and their authors)
25
- Reference source: The pinned responses: repos/dotnet/runtime/issues/49947/events.json and its companion repos/dotnet/runtime/issues/49947/comments.json.
26
  Check: Exact match against values read from the pinned JSON: closed/reopened/merged entries (actor, created_at), assigned/unassigned/review_requested entries (assignee or requested_reviewer, assigner or review_requester, created_at), and comments (author login, created_at). A human comment is a 'commented' entry whose author is not a bot. Bot accounts excluded from the comment count: accounts with type 'Bot', logins ending in '[bot]', and the named automation accounts azure-pipelines, bors, codecov, dotnet-bot, dotnet-issue-labeler, dotnet-maestro, dotnet-policy-service, facebook-github-bot, flutter-dashboard, fluttergithubbot, google-cla, grafanabot, homeassistant, k8s-ci-robot, k8s-triage-robot, linux-foundation-easycla, msftbot, netlify, probot-home-assistant, pytorch-bot, pytorch-probot, rust-highfive, rustbot, typescript-bot, vs-code-engineering, vscode-triage-bot, vscodenpm. Bot accounts seen in this record: azure-pipelines[bot]. Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl -H 'Accept: application/vnd.github+json' https://api.github.com/repos/dotnet/runtime/issues/49947/events?per_page=100 and .../issues/49947/comments?per_page=100, merge by created_at with jq or Python json; 'gh api' reaches the same endpoints
28
  Companions (not included): repos/dotnet/runtime/issues/49947/comments.json
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** curl -H 'Accept: application/vnd.github+json' https://api.github.com/repos/dotnet/runtime/issues/49947/events?per_page=100 and .../issues/49947/comments?per_page=100, merge by created_at with jq or Python json; 'gh api' reaches the same endpoints
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `repos/dotnet/runtime/issues/49947/events.json`: dotnet/runtime pull request #49947: Fix DEBUG/non-DEBUG asm diffs due to instruction group size estimation.
20
  Format: Two GitHub REST API JSON files fetched from the same issue: events.json (GET /repos/{owner}/{repo}/issues/{number}/events, per_page=100: actor, event, created_at, assignee/assigner, requested_reviewer/review_requester, label, commit_id; 25 entries) and the companion comments.json (GET /repos/{owner}/{repo}/issues/{number}/comments, per_page=100: user, created_at, updated_at, author_association, body; 12 entries). Each is one page with fewer than 100 entries, so both are complete. Together they hold the issue timeline without 'cross-referenced' entries, which embed live repository counters and are not byte-stable (see report). Pull request review submissions and inline review comments (pulls/{number}/reviews and pulls/{number}/comments) are not part of this unit.
21
  Origin: Live GitHub REST interface of a closed pull request, collected as a NOASSERTION pointer row under spec general rule 4: the events and comments JSON were each fetched twice apart and were byte-identical. Unit files kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: closed/reopened/merged entries (actor, created_at), assigned/unassigned/review_requested entries (assignee or requested_reviewer, assigner or review_requester, created_at), and comments (author login, created_at). A human comment is a 'commented' entry whose author is not a bot. Bot accounts excluded from the comment count: accounts with type 'Bot', logins ending in '[bot]', and the named automation accounts azure-pipelines, bors, codecov, dotnet-bot, dotnet-issue-labeler, dotnet-maestro, dotnet-policy-service, facebook-github-bot, flutter-dashboard, fluttergithubbot, google-cla, grafanabot, homeassistant, k8s-ci-robot, k8s-triage-robot, linux-foundation-easycla, msftbot, netlify, probot-home-assistant, pytorch-bot, pytorch-probot, rust-highfive, rustbot, typescript-bot, vs-code-engineering, vscode-triage-bot, vscodenpm. Bot accounts seen in this record: azure-pipelines[bot]. Deterministic, no model judgement. Before answers are issued, the extraction program must pass on 2 hand-checked units of this source (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl -H 'Accept: application/vnd.github+json' https://api.github.com/repos/dotnet/runtime/issues/49947/events?per_page=100 and .../issues/49947/comments?per_page=100, merge by created_at with jq or Python json; 'gh api' reaches the same endpoints
24
  Companions (not included): repos/dotnet/runtime/issues/49947/comments.json
campaign_files/2026-10-m-issue-trackers-2/dpdk-users/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the thread headers. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,7 +19,5 @@ Files:
21
  - `m141-05/2019-March.txt.gz`: [dpdk-users] QAT cryto PMD queue pair and Ring Libary (8 messages, 2019-03-07 to 2019-03-12).
22
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
23
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
24
- Reference result: facts computed from the thread headers
25
- Reference source: Message-ID, In-Reply-To, References, From and Date headers of the messages in the thread, read from the pinned month file.
26
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[dpdk-users] Invalid NUMA socket when running helloworld on VM linux' (5 messages, root <fce56772-28bd-4672-8dad-1289f8462c2e@default>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
27
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
 
8
 
9
  **Self-contained:** The file is self-contained.
10
 
 
 
11
  **Tools to open and check:** Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `m141-05/2019-March.txt.gz`: [dpdk-users] QAT cryto PMD queue pair and Ring Libary (8 messages, 2019-03-07 to 2019-03-12).
20
  Format: Pipermail month archive, mbox text gzip-compressed: full headers (From with obfuscated address and name, Date, Subject, Message-ID, In-Reply-To, References) and plain-text bodies. Cut the thread by root Message-ID through the In-Reply-To closure (last References entry when In-Reply-To is missing).
21
  Origin: Real list traffic in a closed past month of the list's Pipermail archive. The unit is one thread, cut out of the month file by its root Message-ID; the month file is the pinned pointer and is delivered whole.
 
 
22
  Check: Exact match against values computed from the pinned month file: parse it as mbox, drop repeated copies of a Message-ID already seen (first copy kept), take the root Message-ID, collect every message whose In-Reply-To (or last References entry when In-Reply-To is missing) chains back to it, then count, take dates in UTC and compare sender names. Message-IDs are compared after removing angle brackets and double quotes. People are counted by person: From entries with the same address, or the same name after removing accents and letter case, are one person (the names in scale are the first spelling of each). The first-reply interval is the exact difference in whole seconds between the Date headers of the root and the earliest reply, converted to UTC; Date headers have one-second resolution, so no rounding is involved. Deterministic. The program was checked on 2 threads of this source by a second, independent parser (splits the raw text on mbox From lines and reads headers with regular expressions, no mailbox or email module): the chosen thread (also read message by message) and '[dpdk-users] Invalid NUMA socket when running helloworld on VM linux' (5 messages, root <fce56772-28bd-4672-8dad-1289f8462c2e@default>); message count, people, last sender and time, first-reply seconds and direct replies agreed on both.
23
  Tools: Read the archive as mbox with Python's mailbox and email modules (or mutt -f / notmuch on the decompressed file); rebuild the thread from Message-ID, In-Reply-To and References headers.
campaign_files/2026-10-m-issue-trackers-2/dragonflybsd-redmine/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
11
- **Answers:** No answer files are included. Reference result: Facts computed from the issue's own journal (status transitions, assignee changes, notes). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** curl the Redmine REST API (/issues/N.json?include=journals) on bugs.dragonflybsd.org and read it with jq or Python json; python-redmine or any Redmine command line client also work.
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,14 +19,10 @@ Files:
21
  - `issues/3131.json`: Dragonflybsd-redmine #3131: master on Ryzen 5 2400G: java freezes system, smp_invltlb 2: waited too long (Resolved, is_closed true, 2020-02-07).
22
  Format: Redmine REST API issue JSON with include=journals: issue fields (status with is_closed, assigned_to, author, created_on, closed_on) and journals[] (user, created_on, notes, details[] with name, old_value, new_value; status_id and assigned_to_id carry numeric ids).
23
  Origin: Live tracker interface of a closed Redmine issue, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
24
- Reference result: facts computed from the issue's own journal (status transitions, assignee changes, notes)
25
- Reference source: The pinned response: issues/3131.json.
26
  Check: Exact match against values read from the pinned JSON: journals[].details[] with name 'status_id' or 'assigned_to_id' (old_value, new_value, journal created_on and user.name) and journals[].notes. Status and user ids are numeric in details; the final status name comes from issue.status (name, is_closed true) and the final assignee name from issue.assigned_to, so questions ask for ids or for the final names only. Deterministic, no model judgement. Human note count excludes bot accounts, recognised by user name against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: no journal user matched. Before answers are issued, the extraction program must pass on 2 hand-checked issues of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
27
  Tools: curl the Redmine REST API (/issues/N.json?include=journals) on bugs.dragonflybsd.org and read it with jq or Python json; python-redmine or any Redmine command line client also work.
28
  - `issues/3237.json`: Dragonflybsd-redmine #3237: msdosfs: can't properly read files longer than couple of kilobytes (Resolved, is_closed true, 2021-01-27).
29
  Format: Redmine REST API issue JSON with include=journals: issue fields (status with is_closed, assigned_to, author, created_on, closed_on) and journals[] (user, created_on, notes, details[] with name, old_value, new_value; status_id and assigned_to_id carry numeric ids).
30
  Origin: Live tracker interface of a closed Redmine issue, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
31
- Reference result: facts computed from the issue's own journal (status transitions, assignee changes, notes)
32
- Reference source: The pinned response: issues/3237.json.
33
  Check: Exact match against values read from the pinned JSON: journals[].details[] with name 'status_id' or 'assigned_to_id' (old_value, new_value, journal created_on and user.name) and journals[].notes. Status and user ids are numeric in details; the final status name comes from issue.status (name, is_closed true) and the final assignee name from issue.assigned_to, so questions ask for ids or for the final names only. Deterministic, no model judgement. Human note count excludes bot accounts, recognised by user name against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: no journal user matched. Before answers are issued, the extraction program must pass on 2 hand-checked issues of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
34
  Tools: curl the Redmine REST API (/issues/N.json?include=journals) on bugs.dragonflybsd.org and read it with jq or Python json; python-redmine or any Redmine command line client also work.
 
8
 
9
  **Self-contained:** Each file is self-contained.
10
 
 
 
11
  **Tools to open and check:** curl the Redmine REST API (/issues/N.json?include=journals) on bugs.dragonflybsd.org and read it with jq or Python json; python-redmine or any Redmine command line client also work.
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `issues/3131.json`: Dragonflybsd-redmine #3131: master on Ryzen 5 2400G: java freezes system, smp_invltlb 2: waited too long (Resolved, is_closed true, 2020-02-07).
20
  Format: Redmine REST API issue JSON with include=journals: issue fields (status with is_closed, assigned_to, author, created_on, closed_on) and journals[] (user, created_on, notes, details[] with name, old_value, new_value; status_id and assigned_to_id carry numeric ids).
21
  Origin: Live tracker interface of a closed Redmine issue, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
22
  Check: Exact match against values read from the pinned JSON: journals[].details[] with name 'status_id' or 'assigned_to_id' (old_value, new_value, journal created_on and user.name) and journals[].notes. Status and user ids are numeric in details; the final status name comes from issue.status (name, is_closed true) and the final assignee name from issue.assigned_to, so questions ask for ids or for the final names only. Deterministic, no model judgement. Human note count excludes bot accounts, recognised by user name against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: no journal user matched. Before answers are issued, the extraction program must pass on 2 hand-checked issues of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
23
  Tools: curl the Redmine REST API (/issues/N.json?include=journals) on bugs.dragonflybsd.org and read it with jq or Python json; python-redmine or any Redmine command line client also work.
24
  - `issues/3237.json`: Dragonflybsd-redmine #3237: msdosfs: can't properly read files longer than couple of kilobytes (Resolved, is_closed true, 2021-01-27).
25
  Format: Redmine REST API issue JSON with include=journals: issue fields (status with is_closed, assigned_to, author, created_on, closed_on) and journals[] (user, created_on, notes, details[] with name, old_value, new_value; status_id and assigned_to_id carry numeric ids).
26
  Origin: Live tracker interface of a closed Redmine issue, collected as a NOASSERTION pointer row under spec general rule 4: two copies fetched apart were byte-identical. Unit file kept only outside the repository for checking and then deleted.
 
 
27
  Check: Exact match against values read from the pinned JSON: journals[].details[] with name 'status_id' or 'assigned_to_id' (old_value, new_value, journal created_on and user.name) and journals[].notes. Status and user ids are numeric in details; the final status name comes from issue.status (name, is_closed true) and the final assignee name from issue.assigned_to, so questions ask for ids or for the final names only. Deterministic, no model judgement. Human note count excludes bot accounts, recognised by user name against the pattern bot|automation|jenkins|gerrit|jira|system|migrat|importer: no journal user matched. Before answers are issued, the extraction program must pass on 2 hand-checked issues of this tracker (spec general rule 5); hand-checking is pending for the delivery step.
28
  Tools: curl the Redmine REST API (/issues/N.json?include=journals) on bugs.dragonflybsd.org and read it with jq or Python json; python-redmine or any Redmine command line client also work.
campaign_files/2026-10-md-media-and-documents/crema-d/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Transcript of a fixed sentence plus emotion labels (intended and crowd-voted). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Rename the file (the sentence code and the intended emotion leak through the name) and give the audio only; task A: transcript, WER against the README sentence; task B: emotion from the voice, scored against the crowd VoiceVote (N) and, separately, the intended label (HAP); report both, since the crowd and the intent disagree on this clip.
14
 
15
  **Licence (SPDX):** ODbL-1.0 AND LicenseRef-OpenDataCommons-DbCL-1.0
@@ -21,7 +19,5 @@ Files:
21
  - `AudioWAV/1015_TIE_HAP_XX.wav`: CREMA-D clip 1015_TIE_HAP_XX: scripted English sentence with intended and crowd-rated emotion.
22
  Format: WAV (16 kHz mono) one acted sentence, Git LFS object served by media.githubusercontent.com
23
  Origin: acted emotional speech recorded in a studio, labels from crowd raters
24
- Reference result: transcript of a fixed sentence plus emotion labels (intended and crowd-voted)
25
- Reference source: sentence: README.md line 164 "- That is exactly what happened (TIE)." (the code TIE in the file name); intended emotion HAP in the file name; crowd vote: processedResults/summaryTable.csv line 1154 "1153","1015_TIE_HAP_XX","N","79.75","H","76","H","69.78" (VoiceVote N at level 79.75 by the column definitions on README.md lines 110 to 113)
26
  Check: rename the file (the sentence code and the intended emotion leak through the name) and give the audio only; task A: transcript, WER against the README sentence; task B: emotion from the voice, scored against the crowd VoiceVote (N) and, separately, the intended label (HAP); report both, since the crowd and the intent disagree on this clip
27
  Companions (not included): README.md; processedResults/summaryTable.csv
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Rename the file (the sentence code and the intended emotion leak through the name) and give the audio only; task A: transcript, WER against the README sentence; task B: emotion from the voice, scored against the crowd VoiceVote (N) and, separately, the intended label (HAP); report both, since the crowd and the intent disagree on this clip.
12
 
13
  **Licence (SPDX):** ODbL-1.0 AND LicenseRef-OpenDataCommons-DbCL-1.0
 
19
  - `AudioWAV/1015_TIE_HAP_XX.wav`: CREMA-D clip 1015_TIE_HAP_XX: scripted English sentence with intended and crowd-rated emotion.
20
  Format: WAV (16 kHz mono) one acted sentence, Git LFS object served by media.githubusercontent.com
21
  Origin: acted emotional speech recorded in a studio, labels from crowd raters
 
 
22
  Check: rename the file (the sentence code and the intended emotion leak through the name) and give the audio only; task A: transcript, WER against the README sentence; task B: emotion from the voice, scored against the crowd VoiceVote (N) and, separately, the intended label (HAP); report both, since the crowd and the intent disagree on this clip
23
  Companions (not included): README.md; processedResults/summaryTable.csv
campaign_files/2026-10-md-media-and-documents/csl-test-suite-date-datebc/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected formatted citation string. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** citeproc command line (citeproc, the Haskell package pandoc uses) or pandoc --citeproc with the CSL style and the item data as CSL JSON; citeproc-js or citeproc-py as libraries
14
 
15
  **Licence (SPDX):** NOASSERTION
@@ -21,8 +19,6 @@ Files:
21
  - `processor-tests/humans/date_DateBC.txt`: CSL processor test "date_DateBC": a year before the common era formatted in a citation.
22
  Format: CSL test fixture text file with sections MODE, RESULT, CSL (a style in XML), INPUT (items as CSL JSON) and VERSION, each between >>===== NAME =====>> and <<===== NAME =====<< lines
23
  Origin: human-written fixture of the official CSL processor test suite (processor-tests/humans), written by the CSL project for processor authors; the pandoc citeproc library runs this suite
24
- Reference result: expected formatted citation string
25
- Reference source: the RESULT section of date_DateBC.txt
26
  Check: show the agent the CSL style and the INPUT items, with MODE citation (hide RESULT); the agent produces the citation string as a CSL processor would; pass when the string equals RESULT exactly
27
  Tools: citeproc command line (citeproc, the Haskell package pandoc uses) or pandoc --citeproc with the CSL style and the item data as CSL JSON; citeproc-js or citeproc-py as libraries
28
  Companions (not included): README.rst
 
8
 
9
  **Self-contained:** The file is self-contained. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** citeproc command line (citeproc, the Haskell package pandoc uses) or pandoc --citeproc with the CSL style and the item data as CSL JSON; citeproc-js or citeproc-py as libraries
12
 
13
  **Licence (SPDX):** NOASSERTION
 
19
  - `processor-tests/humans/date_DateBC.txt`: CSL processor test "date_DateBC": a year before the common era formatted in a citation.
20
  Format: CSL test fixture text file with sections MODE, RESULT, CSL (a style in XML), INPUT (items as CSL JSON) and VERSION, each between >>===== NAME =====>> and <<===== NAME =====<< lines
21
  Origin: human-written fixture of the official CSL processor test suite (processor-tests/humans), written by the CSL project for processor authors; the pandoc citeproc library runs this suite
 
 
22
  Check: show the agent the CSL style and the INPUT items, with MODE citation (hide RESULT); the agent produces the citation string as a CSL processor would; pass when the string equals RESULT exactly
23
  Tools: citeproc command line (citeproc, the Haskell package pandoc uses) or pandoc --citeproc with the CSL style and the item data as CSL JSON; citeproc-js or citeproc-py as libraries
24
  Companions (not included): README.rst
campaign_files/2026-10-md-media-and-documents/csvkit/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected CSV of a typed .xls sheet. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** in2csv -f xls test.xls (csvkit command line); stdout compared with the expected CSV
14
 
15
  **Licence (SPDX):** MIT
@@ -21,8 +19,6 @@ Files:
21
  - `examples/test.xls`: Csvkit in2csv test: legacy Excel sheet with typed columns converted to CSV.
22
  Format: .xls input and one expected CSV file
23
  Origin: a small .xls workbook about Chicago newspapers and its expected CSV made by the csvkit maintainers; tests/test_utilities/test_in2csv.py is the judge
24
- Reference result: expected CSV of a typed .xls sheet
25
- Reference source: examples/testxls_converted.csv
26
  Check: give the agent test.xls and ask for the CSV in2csv writes with format xls; require exact string equality with the expected file (dates, float formatting, booleans and the trimmed extra column must all match)
27
  Tools: in2csv -f xls test.xls (csvkit command line); stdout compared with the expected CSV
28
  Companions (not included): examples/testxls_converted.csv; tests/test_utilities/test_in2csv.py; COPYING
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** in2csv -f xls test.xls (csvkit command line); stdout compared with the expected CSV
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `examples/test.xls`: Csvkit in2csv test: legacy Excel sheet with typed columns converted to CSV.
20
  Format: .xls input and one expected CSV file
21
  Origin: a small .xls workbook about Chicago newspapers and its expected CSV made by the csvkit maintainers; tests/test_utilities/test_in2csv.py is the judge
 
 
22
  Check: give the agent test.xls and ask for the CSV in2csv writes with format xls; require exact string equality with the expected file (dates, float formatting, booleans and the trimmed extra column must all match)
23
  Tools: in2csv -f xls test.xls (csvkit command line); stdout compared with the expected CSV
24
  Companions (not included): examples/testxls_converted.csv; tests/test_utilities/test_in2csv.py; COPYING
campaign_files/2026-10-md-media-and-documents/ctex-org-ctex-kit-xecjk-testfiles/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected normalised TeX log (.tlg) with the box dumps of two pages. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** l3build check basic01 run in xeCJK/ (headless XeLaTeX run, box dump normalised and diffed against basic01.tlg)
14
 
15
  **Licence (SPDX):** LPPL-1.3c
@@ -21,8 +19,6 @@ Files:
21
  - `xeCJK/testfiles/basic01.lvt`: xeCJK regression test basic01: two lines of classical Chinese text and a Chinese-English mixed sentence set in FandolSong with XeLaTeX, checked through the box dump of each page.
22
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
23
  Origin: regression test of the CTeX community maintainers; basic01.tlg is the accepted XeTeX box dump
24
- Reference result: expected normalised TeX log (.tlg) with the box dumps of two pages
25
- Reference source: xeCJK/testfiles/basic01.tlg, saved by the maintainers from a checked run
26
  Check: run basic01.lvt with l3build on XeTeX and compare the normalised .log with basic01.tlg line by line; hide the .tlg when posing the task
27
  Tools: l3build check basic01 run in xeCJK/ (headless XeLaTeX run, box dump normalised and diffed against basic01.tlg)
28
  Companions (not included): xeCJK/testfiles/basic01.tlg; xeCJK/build.lua; README.md
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** l3build check basic01 run in xeCJK/ (headless XeLaTeX run, box dump normalised and diffed against basic01.tlg)
12
 
13
  **Licence (SPDX):** LPPL-1.3c
 
19
  - `xeCJK/testfiles/basic01.lvt`: xeCJK regression test basic01: two lines of classical Chinese text and a Chinese-English mixed sentence set in FandolSong with XeLaTeX, checked through the box dump of each page.
20
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
21
  Origin: regression test of the CTeX community maintainers; basic01.tlg is the accepted XeTeX box dump
 
 
22
  Check: run basic01.lvt with l3build on XeTeX and compare the normalised .log with basic01.tlg line by line; hide the .tlg when posing the task
23
  Tools: l3build check basic01 run in xeCJK/ (headless XeLaTeX run, box dump normalised and diffed against basic01.tlg)
24
  Companions (not included): xeCJK/testfiles/basic01.tlg; xeCJK/build.lua; README.md
campaign_files/2026-10-md-media-and-documents/dcpurton-afesbrand-testfiles/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected normalised TeX log (.tlg) with the box dump of the logo page. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** l3build check afesbrand01 (headless LuaLaTeX run, box dump normalised and diffed against the .tlg)
14
 
15
  **Licence (SPDX):** LPPL-1.3c+
@@ -21,8 +19,6 @@ Files:
21
  - `testfiles/afesbrand01.lvt`: Afesbrand regression test 01: the national logo of a student organisation drawn at 2 cm height on a neutral background, checked through the LuaTeX box dump.
22
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
23
  Origin: regression test of the package author; afesbrand01.tlg is the accepted LuaTeX box dump
24
- Reference result: expected normalised TeX log (.tlg) with the box dump of the logo page
25
- Reference source: testfiles/afesbrand01.tlg, saved by the author from a checked run
26
  Check: run afesbrand01.lvt with l3build on LuaTeX and compare the normalised .log with the .tlg line by line; hide the .tlg when posing the task
27
  Tools: l3build check afesbrand01 (headless LuaLaTeX run, box dump normalised and diffed against the .tlg)
28
  Companions (not included): testfiles/afesbrand01.tlg; build.lua; README.md; COPYING
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** l3build check afesbrand01 (headless LuaLaTeX run, box dump normalised and diffed against the .tlg)
12
 
13
  **Licence (SPDX):** LPPL-1.3c+
 
19
  - `testfiles/afesbrand01.lvt`: Afesbrand regression test 01: the national logo of a student organisation drawn at 2 cm height on a neutral background, checked through the LuaTeX box dump.
20
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
21
  Origin: regression test of the package author; afesbrand01.tlg is the accepted LuaTeX box dump
 
 
22
  Check: run afesbrand01.lvt with l3build on LuaTeX and compare the normalised .log with the .tlg line by line; hide the .tlg when posing the task
23
  Tools: l3build check afesbrand01 (headless LuaLaTeX run, box dump normalised and diffed against the .tlg)
24
  Companions (not included): testfiles/afesbrand01.tlg; build.lua; README.md; COPYING
campaign_files/2026-10-md-media-and-documents/dcpurton-ltx-talk-au-testfiles/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected normalised TeX log (.tlg) with the box dump of the shipped slide. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** l3build check ltx-talk-au-section (headless LuaLaTeX run, two passes, box dump normalised and diffed against the .tlg)
14
 
15
  **Licence (SPDX):** LPPL-1.3c+
@@ -21,8 +19,6 @@ Files:
21
  - `testfiles/ltx-talk-au-section.lvt`: Ltx-talk-au regression test section: a tagged, PDF/UA-2 slide deck with a purple section divider slide, checked through the full box dump of the shipped page.
22
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
23
  Origin: regression test of the class author; the .tlg is the accepted LuaTeX box dump of the divider slide
24
- Reference result: expected normalised TeX log (.tlg) with the box dump of the shipped slide
25
- Reference source: testfiles/ltx-talk-au-section.tlg, saved by the author from a checked run
26
  Check: run the .lvt with l3build on LuaTeX with two runs and compare the normalised .log with the .tlg line by line; hide the .tlg when posing the task
27
  Tools: l3build check ltx-talk-au-section (headless LuaLaTeX run, two passes, box dump normalised and diffed against the .tlg)
28
  Companions (not included): testfiles/ltx-talk-au-section.tlg; build.lua; ltx-talk-au.ins; ltx-talk-au.dtx; COPYING
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** l3build check ltx-talk-au-section (headless LuaLaTeX run, two passes, box dump normalised and diffed against the .tlg)
12
 
13
  **Licence (SPDX):** LPPL-1.3c+
 
19
  - `testfiles/ltx-talk-au-section.lvt`: Ltx-talk-au regression test section: a tagged, PDF/UA-2 slide deck with a purple section divider slide, checked through the full box dump of the shipped page.
20
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
21
  Origin: regression test of the class author; the .tlg is the accepted LuaTeX box dump of the divider slide
 
 
22
  Check: run the .lvt with l3build on LuaTeX with two runs and compare the normalised .log with the .tlg line by line; hide the .tlg when posing the task
23
  Tools: l3build check ltx-talk-au-section (headless LuaLaTeX run, two passes, box dump normalised and diffed against the .tlg)
24
  Companions (not included): testfiles/ltx-talk-au-section.tlg; build.lua; ltx-talk-au.ins; ltx-talk-au.dtx; COPYING
campaign_files/2026-10-md-media-and-documents/dcpurton-sbltex-bibleref-sbl-testfiles/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected normalised TeX log (.tlg) with the box dump of the typeset page. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** l3build check bibleref-sbl-01 run in bibleref-sbl/ (headless pdfLaTeX run, box dump normalised and diffed against the .tlg)
14
 
15
  **Licence (SPDX):** LPPL-1.3c+
@@ -21,8 +19,6 @@ Files:
21
  - `bibleref-sbl/testfiles/bibleref-sbl-01.lvt`: Bibleref-sbl regression test 01: the Old Testament book abbreviations of the SBL Handbook of Style typeset by \bibleverse, including the Septuagint Kingdoms aliases, checked through the box dump.
22
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
23
  Origin: regression test of the package author; bibleref-sbl-01.tlg is the accepted pdfTeX box dump
24
- Reference result: expected normalised TeX log (.tlg) with the box dump of the typeset page
25
- Reference source: bibleref-sbl/testfiles/bibleref-sbl-01.tlg, saved by the author from a checked run
26
  Check: run the .lvt with l3build on pdfTeX and compare the normalised .log with the .tlg line by line; hide the .tlg when posing the task
27
  Tools: l3build check bibleref-sbl-01 run in bibleref-sbl/ (headless pdfLaTeX run, box dump normalised and diffed against the .tlg)
28
  Companions (not included): bibleref-sbl/testfiles/bibleref-sbl-01.tlg; bibleref-sbl/build.lua; build-common.lua; bibleref-sbl/bibleref-sbl.ins; bibleref-sbl/bibleref-sbl.dtx; bibleref-sbl/COPYING
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** l3build check bibleref-sbl-01 run in bibleref-sbl/ (headless pdfLaTeX run, box dump normalised and diffed against the .tlg)
12
 
13
  **Licence (SPDX):** LPPL-1.3c+
 
19
  - `bibleref-sbl/testfiles/bibleref-sbl-01.lvt`: Bibleref-sbl regression test 01: the Old Testament book abbreviations of the SBL Handbook of Style typeset by \bibleverse, including the Septuagint Kingdoms aliases, checked through the box dump.
20
  Format: LaTeX test file (.lvt) plus expected log (.tlg)
21
  Origin: regression test of the package author; bibleref-sbl-01.tlg is the accepted pdfTeX box dump
 
 
22
  Check: run the .lvt with l3build on pdfTeX and compare the normalised .log with the .tlg line by line; hide the .tlg when posing the task
23
  Tools: l3build check bibleref-sbl-01 run in bibleref-sbl/ (headless pdfLaTeX run, box dump normalised and diffed against the .tlg)
24
  Companions (not included): bibleref-sbl/testfiles/bibleref-sbl-01.tlg; bibleref-sbl/build.lua; build-common.lua; bibleref-sbl/bibleref-sbl.ins; bibleref-sbl/bibleref-sbl.dtx; bibleref-sbl/COPYING
campaign_files/2026-10-md-media-and-documents/dherse-codly-tests/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Reference rendering (2 PNG pages) compared pixel by pixel. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** tt run single-line (tytanic test runner, headless Typst compile to PNG and pixel compare); typst compile for a single render
14
 
15
  **Licence (SPDX):** MIT
@@ -21,8 +19,6 @@ Files:
21
  - `tests/single-line/test.typ`: Codly Typst package test single-line: one-line code blocks with line numbers and language icons, with 2 reference PNG pages.
22
  Format: Typst source (.typ) plus reference PNG pages
23
  Origin: visual regression test written by the package authors; ref/*.png are the accepted renderings stored by tytanic
24
- Reference result: reference rendering (2 PNG pages) compared pixel by pixel
25
- Reference source: tests/single-line/ref/1.png and ref/2.png, the renderings the authors accepted
26
  Check: compile test.typ with tytanic 0.2.1 (Typst 0.13.1) and compare each page with ref/N.png pixel by pixel; a pixel deviates when any channel differs by more than max delta; page count must match; hide ref/ when posing the task
27
  Tools: tt run single-line (tytanic test runner, headless Typst compile to PNG and pixel compare); typst compile for a single render
28
  Companions (not included): tests/single-line/ref/1.png; tests/single-line/ref/2.png; codly.typ; src/lib.typ; src/args.typ; src/args.json; src/typst-small.png; typst.toml; LICENSE; .github/workflows/test.yml
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** tt run single-line (tytanic test runner, headless Typst compile to PNG and pixel compare); typst compile for a single render
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `tests/single-line/test.typ`: Codly Typst package test single-line: one-line code blocks with line numbers and language icons, with 2 reference PNG pages.
20
  Format: Typst source (.typ) plus reference PNG pages
21
  Origin: visual regression test written by the package authors; ref/*.png are the accepted renderings stored by tytanic
 
 
22
  Check: compile test.typ with tytanic 0.2.1 (Typst 0.13.1) and compare each page with ref/N.png pixel by pixel; a pixel deviates when any channel differs by more than max delta; page count must match; hide ref/ when posing the task
23
  Tools: tt run single-line (tytanic test runner, headless Typst compile to PNG and pixel compare); typst compile for a single render
24
  Companions (not included): tests/single-line/ref/1.png; tests/single-line/ref/2.png; codly.typ; src/lib.typ; src/args.typ; src/args.json; src/typst-small.png; typst.toml; LICENSE; .github/workflows/test.yml
campaign_files/2026-10-md-media-and-documents/dita-open-toolkit/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected HTML5 files of a filtered DITA map. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** dita --input=all.ditamap --format=html5 --filter=all.ditaval --args.validate=no (DITA-OT command line); output HTML compared with exp/html5 as XML trees
14
 
15
  **Licence (SPDX):** Apache-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `src/test/resources/html5_ditaval/src/all.ditamap`: DITA-OT test: DITA map with DITAVAL passthrough filter published to HTML5.
22
  Format: DITA map, topic and DITAVAL input (3 files) and two expected HTML files
23
  Origin: a DITA map, topic and DITAVAL filter with the expected HTML5 output committed by the DITA-OT maintainers; ITFormatHtml5.html5_ditaval_passthrough_all with AbstractIntegrationTest is the judge
24
- Reference result: expected HTML5 files of a filtered DITA map
25
- Reference source: src/test/resources/html5_ditaval/exp/html5/all.html and exp/html5/index.html
26
  Check: give the agent the three source files and ask for the HTML5 that DITA-OT writes with the DITAVAL filter; parse each produced and expected HTML file and require XML-equal trees; exactly 1 warning is expected
27
  Tools: dita --input=all.ditamap --format=html5 --filter=all.ditaval --args.validate=no (DITA-OT command line); output HTML compared with exp/html5 as XML trees
28
  Companions (not included): src/test/resources/html5_ditaval/src/all.dita; src/test/resources/html5_ditaval/src/all.ditaval; src/test/resources/html5_ditaval/exp/html5/all.html; src/test/resources/html5_ditaval/exp/html5/index.html; src/test/java/org/dita/dost/ITFormatHtml5.java; src/test/java/org/dita/dost/AbstractIntegrationTest.java; LICENSE
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** dita --input=all.ditamap --format=html5 --filter=all.ditaval --args.validate=no (DITA-OT command line); output HTML compared with exp/html5 as XML trees
12
 
13
  **Licence (SPDX):** Apache-2.0
 
19
  - `src/test/resources/html5_ditaval/src/all.ditamap`: DITA-OT test: DITA map with DITAVAL passthrough filter published to HTML5.
20
  Format: DITA map, topic and DITAVAL input (3 files) and two expected HTML files
21
  Origin: a DITA map, topic and DITAVAL filter with the expected HTML5 output committed by the DITA-OT maintainers; ITFormatHtml5.html5_ditaval_passthrough_all with AbstractIntegrationTest is the judge
 
 
22
  Check: give the agent the three source files and ask for the HTML5 that DITA-OT writes with the DITAVAL filter; parse each produced and expected HTML file and require XML-equal trees; exactly 1 warning is expected
23
  Tools: dita --input=all.ditamap --format=html5 --filter=all.ditaval --args.validate=no (DITA-OT command line); output HTML compared with exp/html5 as XML trees
24
  Companions (not included): src/test/resources/html5_ditaval/src/all.dita; src/test/resources/html5_ditaval/src/all.ditaval; src/test/resources/html5_ditaval/exp/html5/all.html; src/test/resources/html5_ditaval/exp/html5/index.html; src/test/java/org/dita/dost/ITFormatHtml5.java; src/test/java/org/dita/dost/AbstractIntegrationTest.java; LICENSE
campaign_files/2026-10-md-media-and-documents/docling-docx-groundtruth/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected Markdown of a Word document (heading, ordered list resumed after a bullet list). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** docling --to md on the .docx (docling CLI) or DocumentConverter().convert(...).document.export_to_markdown(compact_tables=True) in Python; pandoc -t gfm gives a comparable but not identical rendering
14
 
15
  **Licence (SPDX):** MIT
@@ -21,8 +19,6 @@ Files:
21
  - `tests/data/docx/sources/docx_list_resumed_numbering.docx`: Docling Word test: numbered list resumed after an intervening bullet list, converted to Markdown.
22
  Format: Word .docx file plus groundtruth files named <file>.docx.md (also .itxt and .json, not taken)
23
  Origin: test .docx (generated with python-docx, as its core properties say) and groundtruth Markdown written by the docling maintainers; judged by test_e2e_docx_conversions
24
- Reference result: expected Markdown of a Word document (heading, ordered list resumed after a bullet list)
25
- Reference source: tests/data/docx/groundtruth/docx_list_resumed_numbering.docx.md
26
  Check: give the agent the .docx and ask for its Markdown as docling exports it (compact tables); normalise line ends to LF and drop embedded image data on both sides, then compare by exact string equality; any difference fails
27
  Tools: docling --to md on the .docx (docling CLI) or DocumentConverter().convert(...).document.export_to_markdown(compact_tables=True) in Python; pandoc -t gfm gives a comparable but not identical rendering
28
  Companions (not included): tests/data/docx/groundtruth/docx_list_resumed_numbering.docx.md; tests/test_backend_msword.py; tests/verify_utils.py; LICENSE
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** docling --to md on the .docx (docling CLI) or DocumentConverter().convert(...).document.export_to_markdown(compact_tables=True) in Python; pandoc -t gfm gives a comparable but not identical rendering
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `tests/data/docx/sources/docx_list_resumed_numbering.docx`: Docling Word test: numbered list resumed after an intervening bullet list, converted to Markdown.
20
  Format: Word .docx file plus groundtruth files named <file>.docx.md (also .itxt and .json, not taken)
21
  Origin: test .docx (generated with python-docx, as its core properties say) and groundtruth Markdown written by the docling maintainers; judged by test_e2e_docx_conversions
 
 
22
  Check: give the agent the .docx and ask for its Markdown as docling exports it (compact tables); normalise line ends to LF and drop embedded image data on both sides, then compare by exact string equality; any difference fails
23
  Tools: docling --to md on the .docx (docling CLI) or DocumentConverter().convert(...).document.export_to_markdown(compact_tables=True) in Python; pandoc -t gfm gives a comparable but not identical rendering
24
  Companions (not included): tests/data/docx/groundtruth/docx_list_resumed_numbering.docx.md; tests/test_backend_msword.py; tests/verify_utils.py; LICENSE
campaign_files/2026-10-md-media-and-documents/docutils-test-functional/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected LaTeX document. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Hide footnotes_latex.tex; the agent converts footnotes.rst to LaTeX; pass when the output text equals the expected file exactly (test/test_functional.py compare_output, lines 70 to 86).
14
 
15
  **Licence (SPDX):** LicenseRef-docutils-public-domain
@@ -21,8 +19,6 @@ Files:
21
  - `docutils/test/functional/input/footnotes.rst`: Docutils functional test footnotes_latex: reST footnotes and citations converted to LaTeX.
22
  Format: reStructuredText source (49 lines, includes a slice of data/standard.rst) plus the test settings file and the expected LaTeX
23
  Origin: Docutils functional test footnotes_latex: a reStructuredText document of footnotes and citations with the LaTeX the latex2e writer must produce
24
- Reference result: expected LaTeX document
25
- Reference source: docutils/test/functional/expected/footnotes_latex.tex: core.publish_file with writer latex and the settings in footnotes_latex.py
26
  Check: hide footnotes_latex.tex; the agent converts footnotes.rst to LaTeX; pass when the output text equals the expected file exactly (test/test_functional.py compare_output, lines 70 to 86)
27
  Companions (not included): docutils/test/functional/input/data/standard.rst; docutils/test/functional/tests/footnotes_latex.py; docutils/test/functional/expected/footnotes_latex.tex
28
  Dependencies: Docutils
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Hide footnotes_latex.tex; the agent converts footnotes.rst to LaTeX; pass when the output text equals the expected file exactly (test/test_functional.py compare_output, lines 70 to 86).
12
 
13
  **Licence (SPDX):** LicenseRef-docutils-public-domain
 
19
  - `docutils/test/functional/input/footnotes.rst`: Docutils functional test footnotes_latex: reST footnotes and citations converted to LaTeX.
20
  Format: reStructuredText source (49 lines, includes a slice of data/standard.rst) plus the test settings file and the expected LaTeX
21
  Origin: Docutils functional test footnotes_latex: a reStructuredText document of footnotes and citations with the LaTeX the latex2e writer must produce
 
 
22
  Check: hide footnotes_latex.tex; the agent converts footnotes.rst to LaTeX; pass when the output text equals the expected file exactly (test/test_functional.py compare_output, lines 70 to 86)
23
  Companions (not included): docutils/test/functional/input/data/standard.rst; docutils/test/functional/tests/footnotes_latex.py; docutils/test/functional/expected/footnotes_latex.tex
24
  Dependencies: Docutils
campaign_files/2026-10-md-media-and-documents/docutils-test-odt/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected OpenDocument text file (content.xml compared). Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** rst2odt odt_tables1.rst out.odt (docutils odf_odt writer, headless); compare content.xml with Python zipfile and xml.etree; odfpy or soffice --headless to inspect
14
 
15
  **Licence (SPDX):** LicenseRef-docutils-public-domain
@@ -21,8 +19,6 @@ Files:
21
  - `docutils/test/functional/input/odt_tables1.rst`: Docutils ODT writer test odt_tables1: reST grid tables converted to an OpenDocument text file.
22
  Format: reStructuredText source (171 lines) plus the expected OpenDocument text file (.odt, 11250 bytes)
23
  Origin: Docutils ODF writer test odt_tables1: a reStructuredText document of grid and simple tables with the OpenDocument text file the writer must produce
24
- Reference result: expected OpenDocument text file (content.xml compared)
25
- Reference source: docutils/test/functional/expected/odt_tables1.odt: the output of docutils.core.publish_string with odf_odt.Writer() on odt_tables1.rst
26
  Check: hide the .odt; the agent converts the .rst to ODT; unzip both files, read content.xml, sort the attributes of every element and pass when the two serialisations are equal (test/test_writers/test_odt.py lines 62 to 119)
27
  Tools: rst2odt odt_tables1.rst out.odt (docutils odf_odt writer, headless); compare content.xml with Python zipfile and xml.etree; odfpy or soffice --headless to inspect
28
  Companions (not included): docutils/test/functional/expected/odt_tables1.odt
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** rst2odt odt_tables1.rst out.odt (docutils odf_odt writer, headless); compare content.xml with Python zipfile and xml.etree; odfpy or soffice --headless to inspect
12
 
13
  **Licence (SPDX):** LicenseRef-docutils-public-domain
 
19
  - `docutils/test/functional/input/odt_tables1.rst`: Docutils ODT writer test odt_tables1: reST grid tables converted to an OpenDocument text file.
20
  Format: reStructuredText source (171 lines) plus the expected OpenDocument text file (.odt, 11250 bytes)
21
  Origin: Docutils ODF writer test odt_tables1: a reStructuredText document of grid and simple tables with the OpenDocument text file the writer must produce
 
 
22
  Check: hide the .odt; the agent converts the .rst to ODT; unzip both files, read content.xml, sort the attributes of every element and pass when the two serialisations are equal (test/test_writers/test_odt.py lines 62 to 119)
23
  Tools: rst2odt odt_tables1.rst out.odt (docutils odf_odt writer, headless); compare content.xml with Python zipfile and xml.etree; odfpy or soffice --headless to inspect
24
  Companions (not included): docutils/test/functional/expected/odt_tables1.odt
campaign_files/2026-10-md-media-and-documents/docx-preview/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected HTML rendering of a .docx with tables. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** docx-preview renderAsync() run headless (for example in a Karma or Playwright browser driven from the command line); the produced innerHTML compared with result.html after whitespace normalisation
14
 
15
  **Licence (SPDX):** Apache-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `tests/render-test/table/document.docx`: Docx-preview test: Word document with tables rendered to HTML.
22
  Format: .docx input and one expected HTML file
23
  Origin: a Word document with several tables and the expected HTML committed by the docxjs maintainer; tests/render-test/test.spec.js is the judge
24
- Reference result: expected HTML rendering of a .docx with tables
25
- Reference source: tests/render-test/table/result.html
26
  Check: give the agent document.docx and ask for the HTML docx-preview renders with renderChanges on; in both strings turn every run of whitespace into one space and put a newline between adjacent tags, then require exact equality
27
  Tools: docx-preview renderAsync() run headless (for example in a Karma or Playwright browser driven from the command line); the produced innerHTML compared with result.html after whitespace normalisation
28
  Companions (not included): tests/render-test/table/result.html; tests/render-test/test.spec.js; LICENSE
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** docx-preview renderAsync() run headless (for example in a Karma or Playwright browser driven from the command line); the produced innerHTML compared with result.html after whitespace normalisation
12
 
13
  **Licence (SPDX):** Apache-2.0
 
19
  - `tests/render-test/table/document.docx`: Docx-preview test: Word document with tables rendered to HTML.
20
  Format: .docx input and one expected HTML file
21
  Origin: a Word document with several tables and the expected HTML committed by the docxjs maintainer; tests/render-test/test.spec.js is the judge
 
 
22
  Check: give the agent document.docx and ask for the HTML docx-preview renders with renderChanges on; in both strings turn every run of whitespace into one space and put a newline between adjacent tags, then require exact equality
23
  Tools: docx-preview renderAsync() run headless (for example in a Karma or Playwright browser driven from the command line); the produced innerHTML compared with result.html after whitespace normalisation
24
  Companions (not included): tests/render-test/table/result.html; tests/render-test/test.spec.js; LICENSE
campaign_files/2026-10-md-media-and-documents/doxygen/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: Expected Doxygen XML of a documented main page. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** doxygen run on a Doxyfile with GENERATE_XML=YES over 001_a.dox (as testing/runtests.py sets it up); indexpage.xml normalised with xmllint --format --noblanks and compared with the reference
14
 
15
  **Licence (SPDX):** GPL-2.0-only
@@ -21,8 +19,6 @@ Files:
21
  - `testing/001_a.dox`: Doxygen test 001: \a and \mainpage commands converted to XML.
22
  Format: .dox input and one expected XML file
23
  Origin: a .dox test input stating its objective and the file to check, with the reference XML written by the Doxygen maintainers; testing/runtests.py is the judge
24
- Reference result: expected Doxygen XML of a documented main page
25
- Reference source: testing/001/indexpage.xml
26
  Check: give the agent 001_a.dox and ask for the indexpage.xml Doxygen writes; pretty-print with xmllint --format --noblanks, blank the xsd version, split both files at tags, drop empty lines, and require diff -b -w to report no difference
27
  Tools: doxygen run on a Doxyfile with GENERATE_XML=YES over 001_a.dox (as testing/runtests.py sets it up); indexpage.xml normalised with xmllint --format --noblanks and compared with the reference
28
  Companions (not included): testing/001/indexpage.xml; testing/runtests.py; LICENSE
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** doxygen run on a Doxyfile with GENERATE_XML=YES over 001_a.dox (as testing/runtests.py sets it up); indexpage.xml normalised with xmllint --format --noblanks and compared with the reference
12
 
13
  **Licence (SPDX):** GPL-2.0-only
 
19
  - `testing/001_a.dox`: Doxygen test 001: \a and \mainpage commands converted to XML.
20
  Format: .dox input and one expected XML file
21
  Origin: a .dox test input stating its objective and the file to check, with the reference XML written by the Doxygen maintainers; testing/runtests.py is the judge
 
 
22
  Check: give the agent 001_a.dox and ask for the indexpage.xml Doxygen writes; pretty-print with xmllint --format --noblanks, blank the xsd version, split both files at tags, drop empty lines, and require diff -b -w to report no difference
23
  Tools: doxygen run on a Doxyfile with GENERATE_XML=YES over 001_a.dox (as testing/runtests.py sets it up); indexpage.xml normalised with xmllint --format --noblanks and compared with the reference
24
  Companions (not included): testing/001/indexpage.xml; testing/runtests.py; LICENSE
campaign_files/2026-10-ml-serving-configs-2/cpython-cmath-testcases/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_output_values. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Task: implement the named complex function in double precision. Judge as test_specific_values (Lib/test/test_cmath.py from line 280): lines with flag invalid or divide-by-zero expect ValueError, overflow expects OverflowError; otherwise compare the real part and the imaginary part separately with rAssertAlmostEqual (line 69): equal NaNs match, equal infinities match, zeros must match in sign unless the line has ignore-real-sign or ignore-imag-sign, else the difference must be within abs_err or rel_err times the larger magnitude.
14
 
15
  **Licence (SPDX):** PSF-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `Lib/test/mathdata/cmath_testcases.txt`: CPython cmath module test vectors: 2,097 complex cases for acos, acosh, asin, asinh, atan, atanh, cos, cosh, exp, log, log10, sin, sinh, sqrt, tan, tanh, polar and rect, including the C99 Annex G special values, with expected exceptions.
22
  Format: text file, one case per line: id, function name, real and imaginary parts of the argument, '->', real and imaginary parts of the expected value, optional flags (divide-by-zero, invalid, overflow, ignore-real-sign, ignore-imag-sign); '--' comment lines
23
  Origin: test data of the CPython cmath module: reference values and special values for complex elementary functions
24
- Reference result: expected_output_values
25
- Reference source: the expected real and imaginary parts after '->' on each line of Lib/test/mathdata/cmath_testcases.txt, plus the flags column for the expected ValueError or OverflowError and the ignored signs
26
  Check: Task: implement the named complex function in double precision. Judge as test_specific_values (Lib/test/test_cmath.py from line 280): lines with flag invalid or divide-by-zero expect ValueError, overflow expects OverflowError; otherwise compare the real part and the imaginary part separately with rAssertAlmostEqual (line 69): equal NaNs match, equal infinities match, zeros must match in sign unless the line has ignore-real-sign or ignore-imag-sign, else the difference must be within abs_err or rel_err times the larger magnitude
27
  Companions (not included): Lib/test/test_cmath.py; LICENSE
28
  Dependencies: none to read the vectors; to run the original test: CPython test suite at commit 0ec3aee2.
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Task: implement the named complex function in double precision. Judge as test_specific_values (Lib/test/test_cmath.py from line 280): lines with flag invalid or divide-by-zero expect ValueError, overflow expects OverflowError; otherwise compare the real part and the imaginary part separately with rAssertAlmostEqual (line 69): equal NaNs match, equal infinities match, zeros must match in sign unless the line has ignore-real-sign or ignore-imag-sign, else the difference must be within abs_err or rel_err times the larger magnitude.
12
 
13
  **Licence (SPDX):** PSF-2.0
 
19
  - `Lib/test/mathdata/cmath_testcases.txt`: CPython cmath module test vectors: 2,097 complex cases for acos, acosh, asin, asinh, atan, atanh, cos, cosh, exp, log, log10, sin, sinh, sqrt, tan, tanh, polar and rect, including the C99 Annex G special values, with expected exceptions.
20
  Format: text file, one case per line: id, function name, real and imaginary parts of the argument, '->', real and imaginary parts of the expected value, optional flags (divide-by-zero, invalid, overflow, ignore-real-sign, ignore-imag-sign); '--' comment lines
21
  Origin: test data of the CPython cmath module: reference values and special values for complex elementary functions
 
 
22
  Check: Task: implement the named complex function in double precision. Judge as test_specific_values (Lib/test/test_cmath.py from line 280): lines with flag invalid or divide-by-zero expect ValueError, overflow expects OverflowError; otherwise compare the real part and the imaginary part separately with rAssertAlmostEqual (line 69): equal NaNs match, equal infinities match, zeros must match in sign unless the line has ignore-real-sign or ignore-imag-sign, else the difference must be within abs_err or rel_err times the larger magnitude
23
  Companions (not included): Lib/test/test_cmath.py; LICENSE
24
  Dependencies: none to read the vectors; to run the original test: CPython test suite at commit 0ec3aee2.
campaign_files/2026-10-ml-serving-configs-2/cpython-math-testcases/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_output_values. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Task: implement the named special function in double precision. Judge as test_mtestfile (Lib/test/test_math.py lines 2143-2209): for each line call the function on the argument; a line with flag invalid or divide-by-zero expects ValueError, overflow expects OverflowError; otherwise result_check (line 122) passes when the result is within ulp_tol ulps (to_ulps, line 37) or within abs_tol of the expected value (ulp_abs_check, lines 54-70), with the tolerances below.
14
 
15
  **Licence (SPDX):** PSF-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `Lib/test/mathdata/math_testcases.txt`: CPython math module test vectors: 2,001 single-argument cases for erf, erfc, lgamma, gamma, log1p, expm1, log2, sinpi, cospi, tanpi, asinpi, acospi, atanpi and atan2pi with correctly rounded expected values and expected exceptions.
22
  Format: text file, one case per line: id, function name, argument(s) as decimal or hex float literal, '->', expected value, optional flags (divide-by-zero, invalid, overflow, underflow) naming the expected exception; '--' comment lines
23
  Origin: test data of the CPython math module: correctly rounded reference values for libm-style special functions
24
- Reference result: expected_output_values
25
- Reference source: the expected value after '->' on each line of Lib/test/mathdata/math_testcases.txt (header: values computed with MPFR or mpmath to the correctly rounded double), plus the flags column for the expected ValueError or OverflowError
26
  Check: Task: implement the named special function in double precision. Judge as test_mtestfile (Lib/test/test_math.py lines 2143-2209): for each line call the function on the argument; a line with flag invalid or divide-by-zero expects ValueError, overflow expects OverflowError; otherwise result_check (line 122) passes when the result is within ulp_tol ulps (to_ulps, line 37) or within abs_tol of the expected value (ulp_abs_check, lines 54-70), with the tolerances below
27
  Companions (not included): Lib/test/test_math.py; LICENSE
28
  Dependencies: none to read the vectors; to run the original test: CPython test suite at commit 0ec3aee2.
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Task: implement the named special function in double precision. Judge as test_mtestfile (Lib/test/test_math.py lines 2143-2209): for each line call the function on the argument; a line with flag invalid or divide-by-zero expects ValueError, overflow expects OverflowError; otherwise result_check (line 122) passes when the result is within ulp_tol ulps (to_ulps, line 37) or within abs_tol of the expected value (ulp_abs_check, lines 54-70), with the tolerances below.
12
 
13
  **Licence (SPDX):** PSF-2.0
 
19
  - `Lib/test/mathdata/math_testcases.txt`: CPython math module test vectors: 2,001 single-argument cases for erf, erfc, lgamma, gamma, log1p, expm1, log2, sinpi, cospi, tanpi, asinpi, acospi, atanpi and atan2pi with correctly rounded expected values and expected exceptions.
20
  Format: text file, one case per line: id, function name, argument(s) as decimal or hex float literal, '->', expected value, optional flags (divide-by-zero, invalid, overflow, underflow) naming the expected exception; '--' comment lines
21
  Origin: test data of the CPython math module: correctly rounded reference values for libm-style special functions
 
 
22
  Check: Task: implement the named special function in double precision. Judge as test_mtestfile (Lib/test/test_math.py lines 2143-2209): for each line call the function on the argument; a line with flag invalid or divide-by-zero expects ValueError, overflow expects OverflowError; otherwise result_check (line 122) passes when the result is within ulp_tol ulps (to_ulps, line 37) or within abs_tol of the expected value (ulp_abs_check, lines 54-70), with the tolerances below
23
  Companions (not included): Lib/test/test_math.py; LICENSE
24
  Dependencies: none to read the vectors; to run the original test: CPython test suite at commit 0ec3aee2.
campaign_files/2026-10-ml-serving-configs-2/crlibm-testdata/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 11 of 11 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_output_values. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Tools are named in the check written per file in the notes.
14
 
15
  **Licence (SPDX):** GPL-2.0-only
@@ -21,88 +19,66 @@ Files:
21
  - `tests/log.testdata`: CRlibm correctly rounded test vectors for the natural logarithm log (double): 1226 cases with bit-exact expected outputs in the four rounding modes (N 1117, P 28, M 28, Z 53).
22
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
23
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
24
- Reference result: expected_output_values
25
- Reference source: the expected output halves on each line of tests/log.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of log in the line's rounding mode; line 56: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
26
  Check: Task: implement log (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
27
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
28
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
29
  - `tests/log2.testdata`: CRlibm correctly rounded test vectors for the base 2 logarithm log2 (double): 431 cases with bit-exact expected outputs in the four rounding modes (N 104, P 107, M 107, Z 113).
30
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
31
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
32
- Reference result: expected_output_values
33
- Reference source: the expected output halves on each line of tests/log2.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of log2 in the line's rounding mode; line 22: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
34
  Check: Task: implement log2 (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
35
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
36
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
37
  - `tests/log10.testdata`: CRlibm correctly rounded test vectors for the base 10 logarithm log10 (double): 148 cases with bit-exact expected outputs in the four rounding modes (N 52, P 28, M 28, Z 40).
38
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
39
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
40
- Reference result: expected_output_values
41
- Reference source: the expected output halves on each line of tests/log10.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of log10 in the line's rounding mode; line 138: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
42
  Check: Task: implement log10 (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
43
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
44
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
45
  - `tests/log1p.testdata`: CRlibm correctly rounded test vectors for the log1p (log(1+x)) (double): 586 cases with bit-exact expected outputs in the four rounding modes (N 227, P 8, M 8, Z 343).
46
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
47
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
48
- Reference result: expected_output_values
49
- Reference source: the expected output halves on each line of tests/log1p.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of log1p in the line's rounding mode; line 58: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
50
  Check: Task: implement log1p (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
51
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
52
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
53
  - `tests/exp.testdata`: CRlibm correctly rounded test vectors for the exponential exp (double): 6140 cases with bit-exact expected outputs in the four rounding modes (N 4771, P 638, M 13, Z 718).
54
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
55
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
56
- Reference result: expected_output_values
57
- Reference source: the expected output halves on each line of tests/exp.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of exp in the line's rounding mode; line 79: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
58
  Check: Task: implement exp (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
59
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
60
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
61
  - `tests/expm1.testdata`: CRlibm correctly rounded test vectors for the expm1 (exp(x)-1) (double): 732 cases with bit-exact expected outputs in the four rounding modes (N 239, P 34, M 32, Z 427).
62
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
63
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
64
- Reference result: expected_output_values
65
- Reference source: the expected output halves on each line of tests/expm1.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of expm1 in the line's rounding mode; line 188: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
66
  Check: Task: implement expm1 (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
67
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
68
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
69
  - `tests/sinh.testdata`: CRlibm correctly rounded test vectors for the hyperbolic sine sinh (double): 1543 cases with bit-exact expected outputs in the four rounding modes (N 416, P 356, M 356, Z 415).
70
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
71
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
72
- Reference result: expected_output_values
73
- Reference source: the expected output halves on each line of tests/sinh.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of sinh in the line's rounding mode; line 60: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
74
  Check: Task: implement sinh (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
75
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
76
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
77
  - `tests/cosh.testdata`: CRlibm correctly rounded test vectors for the hyperbolic cosine cosh (double): 2105 cases with bit-exact expected outputs in the four rounding modes (N 549, P 495, M 495, Z 566).
78
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
79
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
80
- Reference result: expected_output_values
81
- Reference source: the expected output halves on each line of tests/cosh.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of cosh in the line's rounding mode; line 83: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
82
  Check: Task: implement cosh (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
83
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
84
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
85
  - `tests/asin.testdata`: CRlibm correctly rounded test vectors for the arcsine asin (double): 2625 cases with bit-exact expected outputs in the four rounding modes (N 748, P 649, M 595, Z 633).
86
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
87
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
88
- Reference result: expected_output_values
89
- Reference source: the expected output halves on each line of tests/asin.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of asin in the line's rounding mode; line 158: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
90
  Check: Task: implement asin (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
91
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
92
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
93
  - `tests/atan.testdata`: CRlibm correctly rounded test vectors for the arctangent atan (double): 6852 cases with bit-exact expected outputs in the four rounding modes (N 5623, P 367, M 368, Z 494).
94
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
95
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
96
- Reference result: expected_output_values
97
- Reference source: the expected output halves on each line of tests/atan.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of atan in the line's rounding mode; line 56: "One in five of the very worst cases computed by Lefevre and Muller" (hardest-to-round inputs of the table maker's dilemma); CRlibm is a correctly rounded library and the rig compares bit for bit
98
  Check: Task: implement atan (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
99
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
100
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
101
  - `tests/pow.testdata`: CRlibm correctly rounded test vectors for the power pow(x, y) (double): 10003 cases with bit-exact expected outputs in round to nearest.
102
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves twice (x then y), expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
103
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
104
- Reference result: expected_output_values
105
- Reference source: the expected output halves on each line of tests/pow.testdata (format described on lines 8 to 13 of the file): the correctly rounded double result of pow in the line's rounding mode; CRlibm is a correctly rounded library and the rig compares bit for bit
106
  Check: Task: implement pow (double) with correct rounding to nearest on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81; pow takes two arguments, line 83), read each line as mode, two inputs and the expected result as high and low 32 bit hexadecimal halves (lines 92 to 99), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Every line of this file is mode N; the rig has no directed rounding pow (tests/test_common.c lines 1199 to 1208 set RU, RD and RZ to NULL). Pass when the failure count printed on line 177 is 0
107
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
108
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 11 of 11 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Tools are named in the check written per file in the notes.
12
 
13
  **Licence (SPDX):** GPL-2.0-only
 
19
  - `tests/log.testdata`: CRlibm correctly rounded test vectors for the natural logarithm log (double): 1226 cases with bit-exact expected outputs in the four rounding modes (N 1117, P 28, M 28, Z 53).
20
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
21
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
22
  Check: Task: implement log (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
23
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
24
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
25
  - `tests/log2.testdata`: CRlibm correctly rounded test vectors for the base 2 logarithm log2 (double): 431 cases with bit-exact expected outputs in the four rounding modes (N 104, P 107, M 107, Z 113).
26
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
27
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
28
  Check: Task: implement log2 (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
29
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
30
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
31
  - `tests/log10.testdata`: CRlibm correctly rounded test vectors for the base 10 logarithm log10 (double): 148 cases with bit-exact expected outputs in the four rounding modes (N 52, P 28, M 28, Z 40).
32
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
33
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
34
  Check: Task: implement log10 (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
35
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
36
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
37
  - `tests/log1p.testdata`: CRlibm correctly rounded test vectors for the log1p (log(1+x)) (double): 586 cases with bit-exact expected outputs in the four rounding modes (N 227, P 8, M 8, Z 343).
38
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
39
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
40
  Check: Task: implement log1p (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
41
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
42
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
43
  - `tests/exp.testdata`: CRlibm correctly rounded test vectors for the exponential exp (double): 6140 cases with bit-exact expected outputs in the four rounding modes (N 4771, P 638, M 13, Z 718).
44
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
45
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
46
  Check: Task: implement exp (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
47
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
48
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
49
  - `tests/expm1.testdata`: CRlibm correctly rounded test vectors for the expm1 (exp(x)-1) (double): 732 cases with bit-exact expected outputs in the four rounding modes (N 239, P 34, M 32, Z 427).
50
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
51
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
52
  Check: Task: implement expm1 (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
53
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
54
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
55
  - `tests/sinh.testdata`: CRlibm correctly rounded test vectors for the hyperbolic sine sinh (double): 1543 cases with bit-exact expected outputs in the four rounding modes (N 416, P 356, M 356, Z 415).
56
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
57
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
58
  Check: Task: implement sinh (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
59
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
60
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
61
  - `tests/cosh.testdata`: CRlibm correctly rounded test vectors for the hyperbolic cosine cosh (double): 2105 cases with bit-exact expected outputs in the four rounding modes (N 549, P 495, M 495, Z 566).
62
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
63
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
64
  Check: Task: implement cosh (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
65
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
66
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
67
  - `tests/asin.testdata`: CRlibm correctly rounded test vectors for the arcsine asin (double): 2625 cases with bit-exact expected outputs in the four rounding modes (N 748, P 649, M 595, Z 633).
68
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
69
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
70
  Check: Task: implement asin (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
71
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
72
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
73
  - `tests/atan.testdata`: CRlibm correctly rounded test vectors for the arctangent atan (double): 6852 cases with bit-exact expected outputs in the four rounding modes (N 5623, P 367, M 368, Z 494).
74
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves, expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
75
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
76
  Check: Task: implement atan (double) with correct rounding in each of the four IEEE rounding modes on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81), read each line as rounding mode, input and expected result as high and low 32 bit hexadecimal halves (lines 92 to 99; N, P, M, Z mean to nearest, up, down and toward zero, lines 147 to 150), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Pass when the failure count printed on line 177 is 0
77
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
78
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
79
  - `tests/pow.testdata`: CRlibm correctly rounded test vectors for the power pow(x, y) (double): 10003 cases with bit-exact expected outputs in round to nearest.
80
  Format: text file: first non-comment line is the function name; then one case per line: rounding mode (N, P, M, Z or RN, RU, RD, RZ), input as high and low 32 bit hexadecimal halves twice (x then y), expected output as high and low halves (some lines write the halves with a 0x prefix); '#' starts a comment
81
  Origin: test data of CRlibm, the correctly rounded mathematical library of ENS Lyon (Arenaire project, F. de Dinechin, Ch. Q. Lauter, V. Lefevre, J.-M. Muller), version 1.0beta; fetched from the GitHub mirror taschini/crlibm ("A mirror of the CRLibm project from INRIA Forge"), the original INRIA Forge and lipforge distribution was not fetched
 
 
82
  Check: Task: implement pow (double) with correct rounding to nearest on the raw inputs. Judge as tests/blind_test.c does: read the function name (line 81; pow takes two arguments, line 83), read each line as mode, two inputs and the expected result as high and low 32 bit hexadecimal halves (lines 92 to 99), call the function in that mode and count a failure unless the result bits equal the expected bits exactly; an expected NaN accepts any NaN (lines 133 to 134). Every line of this file is mode N; the rig has no directed rounding pow (tests/test_common.c lines 1199 to 1208 set RU, RD and RZ to NULL). Pass when the failure count printed on line 177 is 0
83
  Companions (not included): tests/blind_test.c; tests/test_common.c; COPYING
84
  Dependencies: none to read the vectors; to run the original rig: CRlibm built from the same commit eb306379 (tests/blind_test.c lines 4-6 include crlibm.h, crlibm_private.h and test_common.h) and optionally MPFR (only to print a reference value on failure, HAVE_MPFR_H).
campaign_files/2026-10-ml-serving-configs-2/ctranslate2-cpu-benchmark/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: aggregate_metric_table. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Translate newstest2014 with each engine and compute type on CPU with the commands in tools/benchmark, score with sacrebleu; BLEU must match each printed value within 0.005 at the pinned versions; tokens per second and memory are description only.
14
 
15
  **Licence (SPDX):** MIT
@@ -21,8 +19,6 @@ Files:
21
  - `ml038-08/README.md`: CTranslate2 README CPU benchmark: BLEU of OpenNMT-tf, OpenNMT-py, Transformers, Marian and CTranslate2 (fp32, int16, int8) on the OpenNMT WMT14 and OPUS-MT models on WMT14 en-de newstest2014, 4 threads on c5.2xlarge.
22
  Format: Markdown README; CPU table lines 79 to 101 with columns tokens per second, max memory and BLEU per engine and compute type
23
  Origin: benchmark results published by the CTranslate2 maintainers.
24
- Reference result: aggregate_metric_table
25
- Reference source: BLEU column of the CPU table (e.g. OpenNMT-py 3.0.4 26.77 at line 86, CTranslate2 3.6.0 26.77 at line 88, int8 and int16 rows below)
26
  Check: Translate newstest2014 with each engine and compute type on CPU with the commands in tools/benchmark, score with sacrebleu; BLEU must match each printed value within 0.005 at the pinned versions; tokens per second and memory are description only
27
  Companions (not included): tools/benchmark/README.md; tools/benchmark/benchmark_all.py; tools/benchmark/opennmt_ende_wmt14/ctranslate2/translate; tools/benchmark/benchmark.py; tools/benchmark/opennmt_ende_wmt14/ctranslate2/Dockerfile; tools/benchmark/opennmt_ende_wmt14/ctranslate2/detokenize; tools/benchmark/opennmt_ende_wmt14/ctranslate2/tokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_py/Dockerfile; tools/benchmark/opennmt_ende_wmt14/opennmt_py/convertv2_v3.py; tools/benchmark/opennmt_ende_wmt14/opennmt_py/detokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_py/tokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_py/translate; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/Dockerfile; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/config.yml; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/detokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/tokenization.yml; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/tokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/translate; tools/benchmark/opus_mt_ende/ctranslate2/Dockerfile; tools/benchmark/opus_mt_ende/ctranslate2/detokenize; tools/benchmark/opus_mt_ende/ctranslate2/tokenize; tools/benchmark/opus_mt_ende/ctranslate2/translate; tools/benchmark/opus_mt_ende/marian/Dockerfile; tools/benchmark/opus_mt_ende/marian/detokenize; tools/benchmark/opus_mt_ende/marian/tokenize; tools/benchmark/opus_mt_ende/marian/translate; tools/benchmark/opus_mt_ende/transformers/Dockerfile; tools/benchmark/opus_mt_ende/transformers/detokenize; tools/benchmark/opus_mt_ende/transformers/tokenize; tools/benchmark/opus_mt_ende/transformers/translate; tools/benchmark/requirements.txt; LICENSE
28
  Dependencies: sacrebleu 1.4.14 (pinned in tools/benchmark/requirements.txt line 3; it also fetches newstest2014 en-de: benchmark_all.py lines 22 to 23 and 142 to 143). docker 4.3.1 and gputil 1.4.0 (tools/benchmark/requirements.txt lines 1 and 2). engines as named in the CPU table, each built from the Dockerfile in its tools/benchmark/<model>/<engine>/ directory (now companions): CTranslate2 3.6.0; OpenNMT-tf 2.31.0 with TensorFlow 2.11.0; OpenNMT-py 3.0.4 with PyTorch 1.13.1; Transformers 4.26.1 with PyTorch 1.13.1; Marian 1.11.0 (README.md lines 84 to 99). OpenNMT WMT14 en-de base Transformer models (OpenNMT-tf and OpenNMT-py releases, README.md lines 68 to 69) and OPUS-MT en-de opus-2020-02-26 (line 70), as downloaded by the Dockerfiles; opus_mt_ende/transformers/Dockerfile line 3 names Helsinki-NLP/opus-mt-en-de without a revision: pinned to Hugging Face revision 6183067f769a302e3861815543b9f312c71b0ca4 (latest commit on main on or before the CTranslate2 commit date 2026-10-05, committed 2023-08-16T11:29:21Z; commits API read 2026-10-06T17:28:47Z). base image of both ctranslate2 Dockerfiles (tools/benchmark/opennmt_ende_wmt14/ctranslate2 and opus_mt_ende/ctranslate2): upstream tag opennmt/ctranslate2:3.6.0-ubuntu22.04-cuda11.2 no longer exists, replaced by opennmt/ctranslate2:3.6.0-ubuntu20.04-cuda11.2@sha256:ec65342844943c5685f4fb48af6484bfcf23f46db3c97d4336afc16737df5880 (checked 2026-10-07: Docker Hub tag list, 406 tags paged, and ghcr.io tag list, 56 tags, each hold one 3.6.0 tag, this one, same digest, pushed 2023-02-16; the 22.04 name is 404 on both); used under its own name, the derived Dockerfile replaces that FROM line; CTranslate2 3.6.0 as the README rows name, OS layer Ubuntu 20.04 (python 3.8) instead of 22.04.
 
8
 
9
  **Self-contained:** Not self-contained: each file is used together with other files from the same upstream. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Translate newstest2014 with each engine and compute type on CPU with the commands in tools/benchmark, score with sacrebleu; BLEU must match each printed value within 0.005 at the pinned versions; tokens per second and memory are description only.
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `ml038-08/README.md`: CTranslate2 README CPU benchmark: BLEU of OpenNMT-tf, OpenNMT-py, Transformers, Marian and CTranslate2 (fp32, int16, int8) on the OpenNMT WMT14 and OPUS-MT models on WMT14 en-de newstest2014, 4 threads on c5.2xlarge.
20
  Format: Markdown README; CPU table lines 79 to 101 with columns tokens per second, max memory and BLEU per engine and compute type
21
  Origin: benchmark results published by the CTranslate2 maintainers.
 
 
22
  Check: Translate newstest2014 with each engine and compute type on CPU with the commands in tools/benchmark, score with sacrebleu; BLEU must match each printed value within 0.005 at the pinned versions; tokens per second and memory are description only
23
  Companions (not included): tools/benchmark/README.md; tools/benchmark/benchmark_all.py; tools/benchmark/opennmt_ende_wmt14/ctranslate2/translate; tools/benchmark/benchmark.py; tools/benchmark/opennmt_ende_wmt14/ctranslate2/Dockerfile; tools/benchmark/opennmt_ende_wmt14/ctranslate2/detokenize; tools/benchmark/opennmt_ende_wmt14/ctranslate2/tokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_py/Dockerfile; tools/benchmark/opennmt_ende_wmt14/opennmt_py/convertv2_v3.py; tools/benchmark/opennmt_ende_wmt14/opennmt_py/detokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_py/tokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_py/translate; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/Dockerfile; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/config.yml; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/detokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/tokenization.yml; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/tokenize; tools/benchmark/opennmt_ende_wmt14/opennmt_tf/translate; tools/benchmark/opus_mt_ende/ctranslate2/Dockerfile; tools/benchmark/opus_mt_ende/ctranslate2/detokenize; tools/benchmark/opus_mt_ende/ctranslate2/tokenize; tools/benchmark/opus_mt_ende/ctranslate2/translate; tools/benchmark/opus_mt_ende/marian/Dockerfile; tools/benchmark/opus_mt_ende/marian/detokenize; tools/benchmark/opus_mt_ende/marian/tokenize; tools/benchmark/opus_mt_ende/marian/translate; tools/benchmark/opus_mt_ende/transformers/Dockerfile; tools/benchmark/opus_mt_ende/transformers/detokenize; tools/benchmark/opus_mt_ende/transformers/tokenize; tools/benchmark/opus_mt_ende/transformers/translate; tools/benchmark/requirements.txt; LICENSE
24
  Dependencies: sacrebleu 1.4.14 (pinned in tools/benchmark/requirements.txt line 3; it also fetches newstest2014 en-de: benchmark_all.py lines 22 to 23 and 142 to 143). docker 4.3.1 and gputil 1.4.0 (tools/benchmark/requirements.txt lines 1 and 2). engines as named in the CPU table, each built from the Dockerfile in its tools/benchmark/<model>/<engine>/ directory (now companions): CTranslate2 3.6.0; OpenNMT-tf 2.31.0 with TensorFlow 2.11.0; OpenNMT-py 3.0.4 with PyTorch 1.13.1; Transformers 4.26.1 with PyTorch 1.13.1; Marian 1.11.0 (README.md lines 84 to 99). OpenNMT WMT14 en-de base Transformer models (OpenNMT-tf and OpenNMT-py releases, README.md lines 68 to 69) and OPUS-MT en-de opus-2020-02-26 (line 70), as downloaded by the Dockerfiles; opus_mt_ende/transformers/Dockerfile line 3 names Helsinki-NLP/opus-mt-en-de without a revision: pinned to Hugging Face revision 6183067f769a302e3861815543b9f312c71b0ca4 (latest commit on main on or before the CTranslate2 commit date 2026-10-05, committed 2023-08-16T11:29:21Z; commits API read 2026-10-06T17:28:47Z). base image of both ctranslate2 Dockerfiles (tools/benchmark/opennmt_ende_wmt14/ctranslate2 and opus_mt_ende/ctranslate2): upstream tag opennmt/ctranslate2:3.6.0-ubuntu22.04-cuda11.2 no longer exists, replaced by opennmt/ctranslate2:3.6.0-ubuntu20.04-cuda11.2@sha256:ec65342844943c5685f4fb48af6484bfcf23f46db3c97d4336afc16737df5880 (checked 2026-10-07: Docker Hub tag list, 406 tags paged, and ghcr.io tag list, 56 tags, each hold one 3.6.0 tag, this one, same digest, pushed 2023-02-16; the 22.04 name is 404 on both); used under its own name, the derived Dockerfile replaces that FROM line; CTranslate2 3.6.0 as the README rows name, OS layer Ubuntu 20.04 (python 3.8) instead of 22.04.
campaign_files/2026-10-ml-serving-configs-2/ctranslate2-engine-test-expected-values/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 3 of 3 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: expected_model_outputs. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** Check: Convert openai/whisper-tiny.en with ctranslate2.converters.TransformersConverter, compute features of jfk.npy with transformers.WhisperProcessor at 16000 Hz, run Whisper.align on CPU with start sequence [50257], the text_tokens and num_frames of the JSON (lines 893 to 903); the alignment list must equal expected_alignments exactly (lines 911 to 913) and the sum of text_token_probs must be within 1e-3 of expected_text_token_probs_sum (lines 907 to 909). For a task, give the audio, the model and the tokens and hide the two expected fields. Load jfk.npy with numpy.load(allow_pickle=False). Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run ctranslate2_test --gtest_filter=*CPU*:OpTest.*; the CPU suites are instantiated at lines 1500 to 1503. Floating-point tests compare with expect_storage_eq at the suite error 1e-5, integer and exact tests with abs_diff 0 (test_utils.h lines 26 to 47 and 63 to 70; at abs_diff 0 floats use EXPECT_FLOAT_EQ); the sampling test Multinomial compares frequencies within 0.05 (line 1078). For a task, give the operator name and inputs of a test and hide its expected literal; the agent computes the outputs. Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run the LayerTest and CPU LayerDeviceFPTest suites (instantiated at lines 261 to 263). Integer position and padder tests compare exactly (expect_storage_eq default abs_diff 0); Alibi uses max(1e-5, 1e-4) (line 37); RotaryEmbedding uses the suite error 1e-5; the position encoder uses 1e-5 (lines 247 and 256). For a task, give the layer and inputs and hide the expected literal.
14
 
15
  **Licence (SPDX):** MIT
@@ -21,24 +19,18 @@ Files:
21
  - `tests/data/audio/jfk_alignments.json`: CTranslate2 Whisper alignment: whisper-tiny.en must align 23 given text tokens of the JFK clip to the published 573 (token, frame) pairs and token-probability sum.
22
  Format: JSON test case (num_frames, text_tokens, expected_text_token_probs_sum, expected_alignments) with the NumPy audio array it belongs to
23
  Origin: expected values written by the CTranslate2 maintainers into the test sources
24
- Reference result: expected_model_outputs
25
- Reference source: expected_alignments and expected_text_token_probs_sum in tests/data/audio/jfk_alignments.json, checked by test_transformers_whisper_align (python/tests/test_transformers.py lines 868 to 915) with the ["jfk"] case
26
  Check: Convert openai/whisper-tiny.en with ctranslate2.converters.TransformersConverter, compute features of jfk.npy with transformers.WhisperProcessor at 16000 Hz, run Whisper.align on CPU with start sequence [50257], the text_tokens and num_frames of the JSON (lines 893 to 903); the alignment list must equal expected_alignments exactly (lines 911 to 913) and the sum of text_token_probs must be within 1e-3 of expected_text_token_probs_sum (lines 907 to 909). For a task, give the audio, the model and the tokens and hide the two expected fields. Load jfk.npy with numpy.load(allow_pickle=False).
27
  Companions (not included): tests/data/audio/jfk.npy; python/tests/test_transformers.py; python/tests/test_utils.py; LICENSE
28
  Dependencies: CTranslate2 (Python package) built from commit 998bb99f7b80e061ff999f69f3764784e265ccbc; transformers and numpy, newest release on or before 2026-10-06 (general rule 1); Hugging Face model openai/whisper-tiny.en revision 87c7102498dcde7456f24cfd30239ca606ed9063 (card license: apache-2.0), not bundled
29
  - `tests/ops_test.cc`: CTranslate2 CPU operators: 87 operator tests (GEMM, softmax, layer and RMS norm, GELU variants, top-k, conv1d, quantization, gather, concat, transpose) must reproduce the published output vectors.
30
  Format: GoogleTest C++ source holding the inputs and expected StorageView values inline, with the comparison helpers header
31
  Origin: expected values written by the CTranslate2 maintainers into the test sources
32
- Reference result: expected_model_outputs
33
- Reference source: the expected StorageView literals inside each test of tests/ops_test.cc, written by the maintainers
34
  Check: Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run ctranslate2_test --gtest_filter=*CPU*:OpTest.*; the CPU suites are instantiated at lines 1500 to 1503. Floating-point tests compare with expect_storage_eq at the suite error 1e-5, integer and exact tests with abs_diff 0 (test_utils.h lines 26 to 47 and 63 to 70; at abs_diff 0 floats use EXPECT_FLOAT_EQ); the sampling test Multinomial compares frequencies within 0.05 (line 1078). For a task, give the operator name and inputs of a test and hide its expected literal; the agent computes the outputs.
35
  Companions (not included): tests/test_utils.h; LICENSE
36
  Dependencies: CTranslate2 source at commit 998bb99f7b80e061ff999f69f3764784e265ccbc built with BUILD_TESTS=ON (CMake, a C++17 compiler, GoogleTest from its third_party submodule, a CPU GEMM backend such as oneDNN, OpenBLAS or Ruy); not bundled
37
  - `tests/layers_test.cc`: CTranslate2 CPU layers: relative positions, ALiBi, rotary embedding, padder and sinusoidal position encoder must reproduce the published values (9 tests).
38
  Format: GoogleTest C++ source holding inputs and expected values inline, with the comparison helpers header
39
  Origin: expected values written by the CTranslate2 maintainers into the test sources
40
- Reference result: expected_model_outputs
41
- Reference source: the expected StorageView literals inside each test of tests/layers_test.cc, written by the maintainers
42
  Check: Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run the LayerTest and CPU LayerDeviceFPTest suites (instantiated at lines 261 to 263). Integer position and padder tests compare exactly (expect_storage_eq default abs_diff 0); Alibi uses max(1e-5, 1e-4) (line 37); RotaryEmbedding uses the suite error 1e-5; the position encoder uses 1e-5 (lines 247 and 256). For a task, give the layer and inputs and hide the expected literal.
43
  Companions (not included): tests/test_utils.h; LICENSE
44
  Dependencies: CTranslate2 source at commit 998bb99f7b80e061ff999f69f3764784e265ccbc built with BUILD_TESTS=ON (CMake, a C++17 compiler, GoogleTest); not bundled
 
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated. 3 of 3 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** Check: Convert openai/whisper-tiny.en with ctranslate2.converters.TransformersConverter, compute features of jfk.npy with transformers.WhisperProcessor at 16000 Hz, run Whisper.align on CPU with start sequence [50257], the text_tokens and num_frames of the JSON (lines 893 to 903); the alignment list must equal expected_alignments exactly (lines 911 to 913) and the sum of text_token_probs must be within 1e-3 of expected_text_token_probs_sum (lines 907 to 909). For a task, give the audio, the model and the tokens and hide the two expected fields. Load jfk.npy with numpy.load(allow_pickle=False). Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run ctranslate2_test --gtest_filter=*CPU*:OpTest.*; the CPU suites are instantiated at lines 1500 to 1503. Floating-point tests compare with expect_storage_eq at the suite error 1e-5, integer and exact tests with abs_diff 0 (test_utils.h lines 26 to 47 and 63 to 70; at abs_diff 0 floats use EXPECT_FLOAT_EQ); the sampling test Multinomial compares frequencies within 0.05 (line 1078). For a task, give the operator name and inputs of a test and hide its expected literal; the agent computes the outputs. Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run the LayerTest and CPU LayerDeviceFPTest suites (instantiated at lines 261 to 263). Integer position and padder tests compare exactly (expect_storage_eq default abs_diff 0); Alibi uses max(1e-5, 1e-4) (line 37); RotaryEmbedding uses the suite error 1e-5; the position encoder uses 1e-5 (lines 247 and 256). For a task, give the layer and inputs and hide the expected literal.
12
 
13
  **Licence (SPDX):** MIT
 
19
  - `tests/data/audio/jfk_alignments.json`: CTranslate2 Whisper alignment: whisper-tiny.en must align 23 given text tokens of the JFK clip to the published 573 (token, frame) pairs and token-probability sum.
20
  Format: JSON test case (num_frames, text_tokens, expected_text_token_probs_sum, expected_alignments) with the NumPy audio array it belongs to
21
  Origin: expected values written by the CTranslate2 maintainers into the test sources
 
 
22
  Check: Convert openai/whisper-tiny.en with ctranslate2.converters.TransformersConverter, compute features of jfk.npy with transformers.WhisperProcessor at 16000 Hz, run Whisper.align on CPU with start sequence [50257], the text_tokens and num_frames of the JSON (lines 893 to 903); the alignment list must equal expected_alignments exactly (lines 911 to 913) and the sum of text_token_probs must be within 1e-3 of expected_text_token_probs_sum (lines 907 to 909). For a task, give the audio, the model and the tokens and hide the two expected fields. Load jfk.npy with numpy.load(allow_pickle=False).
23
  Companions (not included): tests/data/audio/jfk.npy; python/tests/test_transformers.py; python/tests/test_utils.py; LICENSE
24
  Dependencies: CTranslate2 (Python package) built from commit 998bb99f7b80e061ff999f69f3764784e265ccbc; transformers and numpy, newest release on or before 2026-10-06 (general rule 1); Hugging Face model openai/whisper-tiny.en revision 87c7102498dcde7456f24cfd30239ca606ed9063 (card license: apache-2.0), not bundled
25
  - `tests/ops_test.cc`: CTranslate2 CPU operators: 87 operator tests (GEMM, softmax, layer and RMS norm, GELU variants, top-k, conv1d, quantization, gather, concat, transpose) must reproduce the published output vectors.
26
  Format: GoogleTest C++ source holding the inputs and expected StorageView values inline, with the comparison helpers header
27
  Origin: expected values written by the CTranslate2 maintainers into the test sources
 
 
28
  Check: Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run ctranslate2_test --gtest_filter=*CPU*:OpTest.*; the CPU suites are instantiated at lines 1500 to 1503. Floating-point tests compare with expect_storage_eq at the suite error 1e-5, integer and exact tests with abs_diff 0 (test_utils.h lines 26 to 47 and 63 to 70; at abs_diff 0 floats use EXPECT_FLOAT_EQ); the sampling test Multinomial compares frequencies within 0.05 (line 1078). For a task, give the operator name and inputs of a test and hide its expected literal; the agent computes the outputs.
29
  Companions (not included): tests/test_utils.h; LICENSE
30
  Dependencies: CTranslate2 source at commit 998bb99f7b80e061ff999f69f3764784e265ccbc built with BUILD_TESTS=ON (CMake, a C++17 compiler, GoogleTest from its third_party submodule, a CPU GEMM backend such as oneDNN, OpenBLAS or Ruy); not bundled
31
  - `tests/layers_test.cc`: CTranslate2 CPU layers: relative positions, ALiBi, rotary embedding, padder and sinusoidal position encoder must reproduce the published values (9 tests).
32
  Format: GoogleTest C++ source holding inputs and expected values inline, with the comparison helpers header
33
  Origin: expected values written by the CTranslate2 maintainers into the test sources
 
 
34
  Check: Build CTranslate2 at the pinned commit with -DBUILD_TESTS=ON and run the LayerTest and CPU LayerDeviceFPTest suites (instantiated at lines 261 to 263). Integer position and padder tests compare exactly (expect_storage_eq default abs_diff 0); Alibi uses max(1e-5, 1e-4) (line 37); RotaryEmbedding uses the suite error 1e-5; the position encoder uses 1e-5 (lines 247 and 256). For a task, give the layer and inputs and hide the expected literal.
35
  Companions (not included): tests/test_utils.h; LICENSE
36
  Dependencies: CTranslate2 source at commit 998bb99f7b80e061ff999f69f3764784e265ccbc built with BUILD_TESTS=ON (CMake, a C++17 compiler, GoogleTest); not bundled
campaign_files/2026-10-op-rule-calculations-2/dbt-jaffle-shop-seeds/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: computed_from_unit. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** python csv module or pandas; sqlite3 .import or duckdb read_csv; dbt run --select orders with the DuckDB adapter reproduces the reference model
14
 
15
  **Licence (SPDX):** Apache-2.0
@@ -21,8 +19,6 @@ Files:
21
  - `seeds/raw_orders.csv`: jaffle_shop seed tables: orders, payments in cents and customers; compute order amounts by payment method and in dollars.
22
  Format: CSV seeds (3 files, 6,585 bytes) plus 2 dbt SQL models
23
  Origin: demo seed data of the dbt tutorial project; made up
24
- Reference result: computed_from_unit
25
- Reference source: models/staging/stg_payments.sql line 19 (amount / 100 as amount, cents to dollars) and models/orders.sql lines 21 and 24 (sum per payment method, total_amount per order); e.g. order 1 amount 10 (raw_payments.csv line 2: 1000 cents credit_card), order 9 amount 23 (line 10: 2300 cents gift_card plus a 0 bank_transfer)
26
  Check: ask for per-order amounts in dollars, per-method splits or totals by status; compare with the values recomputed by the model rule, exact at cent precision
27
  Tools: python csv module or pandas; sqlite3 .import or duckdb read_csv; dbt run --select orders with the DuckDB adapter reproduces the reference model
28
  Companions (not included): seeds/raw_payments.csv; seeds/raw_customers.csv; models/staging/stg_payments.sql; models/orders.sql
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** python csv module or pandas; sqlite3 .import or duckdb read_csv; dbt run --select orders with the DuckDB adapter reproduces the reference model
12
 
13
  **Licence (SPDX):** Apache-2.0
 
19
  - `seeds/raw_orders.csv`: jaffle_shop seed tables: orders, payments in cents and customers; compute order amounts by payment method and in dollars.
20
  Format: CSV seeds (3 files, 6,585 bytes) plus 2 dbt SQL models
21
  Origin: demo seed data of the dbt tutorial project; made up
 
 
22
  Check: ask for per-order amounts in dollars, per-method splits or totals by status; compare with the values recomputed by the model rule, exact at cent precision
23
  Tools: python csv module or pandas; sqlite3 .import or duckdb read_csv; dbt run --select orders with the DuckDB adapter reproduces the reference model
24
  Companions (not included): seeds/raw_payments.csv; seeds/raw_customers.csv; models/staging/stg_payments.sql; models/orders.sql
campaign_files/2026-10-op-rule-calculations-2/dib-2019-sandwich-analytics/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** The file is self-contained. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: published_metric_values. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** read the CSV with Python csv, compute mean and sample standard deviation of Price, Sold, Del, Sales, Tot_Sls, Unit_Cost, Cost, Cost_Percent, Margin and Profit over all rows; write them to JSON
14
 
15
  **Licence (SPDX):** CC-BY-4.0
@@ -21,8 +19,6 @@ Files:
21
  - `op120-04/WEEKLY_SALES_10STORES.csv`: Sandwich analytics: one year of weekly item-level point-of-sale data from 10 quick-serve sandwich stores, with the published means and standard deviations of the sales table.
22
  Format: CSV, 79,459 rows plus header, 7,868,054 bytes; companions: the article full text (JATS XML) and the Mendeley dataset record (JSON, with the licence)
23
  Origin: article: "The primary sales data and all secondary data are genuine", from the POS system of a US sandwich chain
24
- Reference result: published_metric_values
25
- Reference source: PMC6723374_fullTextXML.xml, Table 1 "Data descriptions, averages, and standard deviations for the Weekly Sales Table": Mean (SD) for 10 numeric columns, e.g. Price "4.50 (3.42)", Sales "61.89 (137.41)", Profit "42.39 (94.47)"
26
  Check: give the CSV; ask for the mean and standard deviation of each numeric sales column over all rows; compare each with Table 1 to its printed precision (two decimals)
27
  Tools: read the CSV with Python csv, compute mean and sample standard deviation of Price, Sold, Del, Sales, Tot_Sls, Unit_Cost, Cost, Cost_Percent, Margin and Profit over all rows; write them to JSON
28
  Companions (not included): PMC6723374_fullTextXML.xml; mendeley_dataset_6htjnfs78b.json
 
8
 
9
  **Self-contained:** The file is self-contained. 1 of 1 files name companion files (other upstream files read alongside); companions are not included in this delivery and are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** read the CSV with Python csv, compute mean and sample standard deviation of Price, Sold, Del, Sales, Tot_Sls, Unit_Cost, Cost, Cost_Percent, Margin and Profit over all rows; write them to JSON
12
 
13
  **Licence (SPDX):** CC-BY-4.0
 
19
  - `op120-04/WEEKLY_SALES_10STORES.csv`: Sandwich analytics: one year of weekly item-level point-of-sale data from 10 quick-serve sandwich stores, with the published means and standard deviations of the sales table.
20
  Format: CSV, 79,459 rows plus header, 7,868,054 bytes; companions: the article full text (JATS XML) and the Mendeley dataset record (JSON, with the licence)
21
  Origin: article: "The primary sales data and all secondary data are genuine", from the POS system of a US sandwich chain
 
 
22
  Check: give the CSV; ask for the mean and standard deviation of each numeric sales column over all rows; compare each with Table 1 to its printed precision (two decimals)
23
  Tools: read the CSV with Python csv, compute mean and sample standard deviation of Price, Sold, Del, Sales, Tot_Sls, Unit_Cost, Cost, Cost_Percent, Margin and Profit over all rows; write them to JSON
24
  Companions (not included): PMC6723374_fullTextXML.xml; mendeley_dataset_6htjnfs78b.json
campaign_files/2026-10-op-rule-calculations-2/dol-whd-flsa-opinion-letters/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated.
10
 
11
- **Answers:** No answer files are included. Reference result: published_adjudication_labels. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
14
 
15
  **Licence (SPDX):** LicenseRef-US-Government-Work
@@ -20,17 +18,13 @@ Files:
20
 
21
  - `sites/dolgov/files/WHD/opinion-letters/FLSA/2020_06_25_06_FLSA.pdf`: WHD opinion letter FLSA2020-6 (June 25, 2020): whether salespeople of the requester meet the outside sales exemption.
22
  Format: PDF, 323467 bytes; text layer extracts with pypdf
23
- Reference result: published_adjudication_labels
24
  Check: give the agent the request paragraph with its last sentence removed when it states the conclusion, and the BACKGROUND (facts presented); hide the analysis section (every sentence that applies the regulation to the facts) and the CONCLUSION section. The agent answers exempt or not exempt; pass when every label equals the conclusion. Exact label match, no tolerance; a conclusion stated with likely counts as its plain label
25
  Tools: extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
26
  - `sites/dolgov/files/WHD/opinion-letters/FLSA/2020_11_30_18_FLSA.pdf`: WHD opinion letter FLSA2020-18 (November 30, 2020): whether workers of an insect farm (superworms for animal feed, crickets and discoid roaches for human food) fall under the agricultural employee exemption from overtime pay.
27
  Format: PDF, 253434 bytes; AES-encrypted with an empty user password (owner restrictions only): pypdf without the cryptography package cannot read it, macOS PDFKit extracts the text
28
- Reference result: published_adjudication_labels
29
- Reference source: the CONCLUSION section: "For the foregoing reasons, all of the employees you have mentioned as working for your insect farm are exempt from the overtime pay provisions of the FLSA."
30
  Check: give the agent the request paragraph with its last sentence removed when it states the conclusion, and the BACKGROUND (facts presented); hide the analysis section (every sentence that applies the regulation to the facts) and the CONCLUSION section. The agent answers exempt or not exempt; pass when every label equals the conclusion. Exact label match, no tolerance; a conclusion stated with likely counts as its plain label
31
  Tools: extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
32
  - `sites/dolgov/files/WHD/opinion-letters/FLSA/2020_08_31_13_FLSA.pdf`: WHD opinion letter FLSA2020-13 (August 31, 2020): part-time corporate-management trainers paid a day rate plus hourly pay: learned professional duties, salary basis, effect of development-work payments, and the highly compensated employee test.
33
  Format: PDF, 290564 bytes; AES-encrypted with an empty user password (owner restrictions only): pypdf without the cryptography package cannot read it, macOS PDFKit extracts the text
34
- Reference result: published_adjudication_labels
35
  Check: give the agent the request paragraph with its last sentence removed when it states the conclusion, and the BACKGROUND (facts presented); hide the analysis section (every sentence that applies the regulation to the facts) and the CONCLUSION section. The agent answers four yes/no questions: (1) do the employees perform learned professional duties (yes), (2) do the delivery-work payments satisfy the salary basis test (no), (3) would the development-work payments cause loss of an otherwise-applicable professional exemption (no), (4) can the highly compensated employee test be satisfied by payments proportional to work performed by a part-time employee (no); pass when every label equals the conclusion. Exact label match, no tolerance; a conclusion stated with likely counts as its plain label
36
  Tools: extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
 
8
 
9
  **Self-contained:** Whether each file needs other upstream files is not stated.
10
 
 
 
11
  **Tools to open and check:** extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
12
 
13
  **Licence (SPDX):** LicenseRef-US-Government-Work
 
18
 
19
  - `sites/dolgov/files/WHD/opinion-letters/FLSA/2020_06_25_06_FLSA.pdf`: WHD opinion letter FLSA2020-6 (June 25, 2020): whether salespeople of the requester meet the outside sales exemption.
20
  Format: PDF, 323467 bytes; text layer extracts with pypdf
 
21
  Check: give the agent the request paragraph with its last sentence removed when it states the conclusion, and the BACKGROUND (facts presented); hide the analysis section (every sentence that applies the regulation to the facts) and the CONCLUSION section. The agent answers exempt or not exempt; pass when every label equals the conclusion. Exact label match, no tolerance; a conclusion stated with likely counts as its plain label
22
  Tools: extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
23
  - `sites/dolgov/files/WHD/opinion-letters/FLSA/2020_11_30_18_FLSA.pdf`: WHD opinion letter FLSA2020-18 (November 30, 2020): whether workers of an insect farm (superworms for animal feed, crickets and discoid roaches for human food) fall under the agricultural employee exemption from overtime pay.
24
  Format: PDF, 253434 bytes; AES-encrypted with an empty user password (owner restrictions only): pypdf without the cryptography package cannot read it, macOS PDFKit extracts the text
 
 
25
  Check: give the agent the request paragraph with its last sentence removed when it states the conclusion, and the BACKGROUND (facts presented); hide the analysis section (every sentence that applies the regulation to the facts) and the CONCLUSION section. The agent answers exempt or not exempt; pass when every label equals the conclusion. Exact label match, no tolerance; a conclusion stated with likely counts as its plain label
26
  Tools: extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
27
  - `sites/dolgov/files/WHD/opinion-letters/FLSA/2020_08_31_13_FLSA.pdf`: WHD opinion letter FLSA2020-13 (August 31, 2020): part-time corporate-management trainers paid a day rate plus hourly pay: learned professional duties, salary basis, effect of development-work payments, and the highly compensated employee test.
28
  Format: PDF, 290564 bytes; AES-encrypted with an empty user password (owner restrictions only): pypdf without the cryptography package cannot read it, macOS PDFKit extracts the text
 
29
  Check: give the agent the request paragraph with its last sentence removed when it states the conclusion, and the BACKGROUND (facts presented); hide the analysis section (every sentence that applies the regulation to the facts) and the CONCLUSION section. The agent answers four yes/no questions: (1) do the employees perform learned professional duties (yes), (2) do the delivery-work payments satisfy the salary basis test (no), (3) would the development-work payments cause loss of an otherwise-applicable professional exemption (no), (4) can the highly compensated employee test be satisfied by payments proportional to work performed by a part-time employee (no); pass when every label equals the conclusion. Exact label match, no tolerance; a conclusion stated with likely counts as its plain label
30
  Tools: extract the text with pypdf (or pdftotext); the encrypted files need a reader that handles AES with an empty user password (pypdf with the cryptography package, qpdf --decrypt, or macOS PDFKit); split at the headings BACKGROUND, the analysis and CONCLUSION; the agent writes the labels to a JSON file
campaign_files/2026-10-op-rule-calculations-2/dolibarr-phpunit-stock/README.md CHANGED
@@ -8,8 +8,6 @@
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
11
- **Answers:** No answer files are included. Reference result: test_assertions. Where the reference result comes from and how a solution is checked is written per file in the notes.
12
-
13
  **Tools to open and check:** phpunit on test/phpunit/MouvementStockTest.php (headless, against a MySQL or PostgreSQL test database), or the Dolibarr REST API to create stock movements and read product stock
14
 
15
  **Licence (SPDX):** GPL-3.0-or-later
@@ -21,8 +19,6 @@ Files:
21
  - `test/phpunit/MouvementStockTest.php`: Dolibarr stock movement test: stock on hand across two warehouses, weighted average cost (PMP) and fractional quantities.
22
  Format: PHPUnit test class, 284 lines
23
  Origin: test written by the project authors; products, quantities and prices are made up
24
- Reference result: test_assertions
25
- Reference source: stock_reel 28 (line 241), PMP 9.8 (line 244) and fractional balance 2.5 (line 282)
26
  Check: give the in and out movements per warehouse with unit prices and ask the total physical stock, the weighted average cost and the balance after a fractional movement; compare as PHPUnit assertEquals does
27
  Tools: phpunit on test/phpunit/MouvementStockTest.php (headless, against a MySQL or PostgreSQL test database), or the Dolibarr REST API to create stock movements and read product stock
28
  Dependencies: Dolibarr at commit ba915623d3e6aa6d57ae3b7d954b0fa5c1b645e6, PHP 8, PHPUnit, MySQL/MariaDB or PostgreSQL test database.
 
8
 
9
  **Self-contained:** Whether the file needs other upstream files is not stated. Dependencies named upstream are listed per file in the notes.
10
 
 
 
11
  **Tools to open and check:** phpunit on test/phpunit/MouvementStockTest.php (headless, against a MySQL or PostgreSQL test database), or the Dolibarr REST API to create stock movements and read product stock
12
 
13
  **Licence (SPDX):** GPL-3.0-or-later
 
19
  - `test/phpunit/MouvementStockTest.php`: Dolibarr stock movement test: stock on hand across two warehouses, weighted average cost (PMP) and fractional quantities.
20
  Format: PHPUnit test class, 284 lines
21
  Origin: test written by the project authors; products, quantities and prices are made up
 
 
22
  Check: give the in and out movements per warehouse with unit prices and ask the total physical stock, the weighted average cost and the balance after a fractional movement; compare as PHPUnit assertEquals does
23
  Tools: phpunit on test/phpunit/MouvementStockTest.php (headless, against a MySQL or PostgreSQL test database), or the Dolibarr REST API to create stock movements and read product stock
24
  Dependencies: Dolibarr at commit ba915623d3e6aa6d57ae3b7d954b0fa5c1b645e6, PHP 8, PHPUnit, MySQL/MariaDB or PostgreSQL test database.