nmarshall commited on
Commit
73a5d10
·
1 Parent(s): 73c9142

qvm: contact basis and damage dispatch

Browse files
docs/ASSET_PIPELINE_TODO.md CHANGED
@@ -799,3 +799,9 @@ Research progress: RE-710 [report](research/QVM_DAMAGE_INTERRUPT.md); remaining
799
 
800
 
801
  Research progress: RE-711 [report](research/QVM_DAMAGE_COMPLETE.md); remaining native/content/runtime/platform scope stays open.
 
 
 
 
 
 
 
799
 
800
 
801
  Research progress: RE-711 [report](research/QVM_DAMAGE_COMPLETE.md); remaining native/content/runtime/platform scope stays open.
802
+
803
+
804
+ Research progress: RE-712 [report](research/QVM_CONTACT_DISPATCH.md); remaining native/content/runtime/platform scope stays open.
805
+
806
+
807
+ RE-703–712 batch complete: [damage factor and contact dispatch](research/QVM_DAMAGE_FACTOR_BATCH.md). Complete289594 handles original-model contact factors and wound bookkeeping;211239 is still captured. RE-702 is partially advanced and RE-713 tracks contact/event/health composition, native parity and Rust/Bevy/Solari platform work. No playable-game milestone is claimed.
docs/FINDINGS.md CHANGED
@@ -1145,3 +1145,9 @@ Bounded qagame paths recover bomb-plant/inactivity correction requests, direct f
1145
 
1146
 
1147
  **RE-711 — Complete damage factor on original-model contacts:** Complete289594 now runs original-model hit contexts through actual inventory, factor selection, wound bookkeeping and every numeric return. Its weapon4 cubic local is computed but never used to scale the return. [Evidence](research/QVM_DAMAGE_COMPLETE.md).
 
 
 
 
 
 
 
1145
 
1146
 
1147
  **RE-711 — Complete damage factor on original-model contacts:** Complete289594 now runs original-model hit contexts through actual inventory, factor selection, wound bookkeeping and every numeric return. Its weapon4 cubic local is computed but never used to scale the return. [Evidence](research/QVM_DAMAGE_COMPLETE.md).
1148
+
1149
+
1150
+ **RE-712 — Contact direction basis and health-damage dispatch:** Contact293373 recomputes global forward/right/up from attacker angles, calls211239 only for nonzero scalar and nonspecial contact, truncates scalar to integer damage, and returns the original float scalar. [Evidence](research/QVM_CONTACT_DISPATCH.md).
1151
+
1152
+
1153
+ RE-703–712 batch complete: [damage factor and contact dispatch](research/QVM_DAMAGE_FACTOR_BATCH.md). Complete289594 handles original-model contact factors and wound bookkeeping;211239 is still captured. RE-702 is partially advanced and RE-713 tracks contact/event/health composition, native parity and Rust/Bevy/Solari platform work. No playable-game milestone is claimed.
docs/PROVISIONAL_TODO.md CHANGED
@@ -1134,3 +1134,9 @@ Research progress: RE-710 [report](research/QVM_DAMAGE_INTERRUPT.md); remaining
1134
 
1135
 
1136
  Research progress: RE-711 [report](research/QVM_DAMAGE_COMPLETE.md); remaining native/content/runtime/platform scope stays open.
 
 
 
 
 
 
 
1134
 
1135
 
1136
  Research progress: RE-711 [report](research/QVM_DAMAGE_COMPLETE.md); remaining native/content/runtime/platform scope stays open.
1137
+
1138
+
1139
+ Research progress: RE-712 [report](research/QVM_CONTACT_DISPATCH.md); remaining native/content/runtime/platform scope stays open.
1140
+
1141
+
1142
+ RE-703–712 batch complete: [damage factor and contact dispatch](research/QVM_DAMAGE_FACTOR_BATCH.md). Complete289594 handles original-model contact factors and wound bookkeeping;211239 is still captured. RE-702 is partially advanced and RE-713 tracks contact/event/health composition, native parity and Rust/Bevy/Solari platform work. No playable-game milestone is claimed.
docs/RESEARCH_LEDGER.md CHANGED
@@ -1197,3 +1197,11 @@ Held item low-byte14 at factor≥0.2 or23 at factor≥0.5 writes client6304=time
1197
  ## 2026-10-21 — RE-711 Complete damage factor on original-model contacts (separate task commit)
1198
 
1199
  Complete289594 now runs original-model hit contexts through actual inventory, factor selection, wound bookkeeping and every numeric return. Its weapon4 cubic local is computed but never used to scale the return. 144 complete-call probes pass a separate source-order float32 and all-external-field oracle, with exact original-context report identity. Full suite: 1296 tests pass. Report regenerates byte-identically (487781 bytes). Storage/filter, links and whitespace checks pass. No assets/Rust runtime/shared VM evaluator changed; native/reference/legal/full platform scope remains open.
 
 
 
 
 
 
 
 
 
1197
  ## 2026-10-21 — RE-711 Complete damage factor on original-model contacts (separate task commit)
1198
 
1199
  Complete289594 now runs original-model hit contexts through actual inventory, factor selection, wound bookkeeping and every numeric return. Its weapon4 cubic local is computed but never used to scale the return. 144 complete-call probes pass a separate source-order float32 and all-external-field oracle, with exact original-context report identity. Full suite: 1296 tests pass. Report regenerates byte-identically (487781 bytes). Storage/filter, links and whitespace checks pass. No assets/Rust runtime/shared VM evaluator changed; native/reference/legal/full platform scope remains open.
1200
+
1201
+
1202
+ ## 2026-10-21 — RE-712 Contact direction basis and health-damage dispatch (separate task commit)
1203
+
1204
+ Contact293373 recomputes global forward/right/up from attacker angles, calls211239 only for nonzero scalar and nonspecial contact, truncates scalar to integer damage, and returns the original float scalar. 120 angle/weapon/scalar/special fixtures verify exact source-order basis words, signed truncation, all nine request arguments and original nearest-point provenance. Full suite: 1300 tests pass. Report regenerates byte-identically (547093 bytes). Storage/filter, links and whitespace checks pass. No assets/Rust runtime/shared VM evaluator changed; native/reference/legal/full platform scope remains open.
1205
+
1206
+
1207
+ Batch RE-703–712 closes ten independently checked research tasks in ten separate commits, advancing the Python baseline1259→1300. All ten reports regenerate byte-identically below2MiB. See the [batch overview](research/QVM_DAMAGE_FACTOR_BATCH.md) for findings, native-model boundaries and remaining work. Original assets, Rust code and the shared VM evaluator are unchanged. RE-702 remains partially open; RE-713 records the next scope.
docs/REVERSE_ENGINEERING_TODO.md CHANGED
@@ -1149,7 +1149,7 @@ Purpose: turn the supplied installation into an evidence-backed specification of
1149
  - [x] RE-701 Rust original-byte MD3 asset core and cross-language validation ([report](research/RUST_MD3.md)).
1150
 
1151
 
1152
- - [ ] RE-702 Integrate the verified Rust MD3 core with Bevy, port original transform/geometry contracts, recover native loader/reference behavior and complete contact/damage/shared-pellet callers toward a playable macOS/Linux Solari slice. See the [next ten tasks](research/ORIGINAL_MODELS_RUST_BATCH.md#remaining-limits-and-next-ten-tasks).
1153
 
1154
 
1155
  - [x] RE-703 Damage protection and armor location selection ([report](research/QVM_DAMAGE_PROTECTION.md)).
@@ -1177,3 +1177,9 @@ Purpose: turn the supplied installation into an evidence-backed specification of
1177
 
1178
 
1179
  - [x] RE-711 Complete damage factor on original-model contacts ([report](research/QVM_DAMAGE_COMPLETE.md)).
 
 
 
 
 
 
 
1149
  - [x] RE-701 Rust original-byte MD3 asset core and cross-language validation ([report](research/RUST_MD3.md)).
1150
 
1151
 
1152
+ - [ ] RE-702 Integrate the verified Rust MD3 core with Bevy, port original transform/geometry contracts, recover native loader/reference behavior and complete contact/damage/shared-pellet callers toward a playable macOS/Linux Solari slice. See the [next ten tasks](research/ORIGINAL_MODELS_RUST_BATCH.md#remaining-limits-and-next-ten-tasks). Partially advanced by RE-703–712: complete289594 factor/wound logic and the contact damage-dispatch suffix are recovered; health mutations, complete contact composition and platform work remain open.
1153
 
1154
 
1155
  - [x] RE-703 Damage protection and armor location selection ([report](research/QVM_DAMAGE_PROTECTION.md)).
 
1177
 
1178
 
1179
  - [x] RE-711 Complete damage factor on original-model contacts ([report](research/QVM_DAMAGE_COMPLETE.md)).
1180
+
1181
+
1182
+ - [x] RE-712 Contact direction basis and health-damage dispatch ([report](research/QVM_CONTACT_DISPATCH.md)).
1183
+
1184
+
1185
+ - [ ] RE-713 Compose complete contact292545 and health211239, continue shared/pellet callers, capture native parity, and integrate Rust/Bevy/Solari toward playable macOS/Linux builds. See the [next ten tasks](research/QVM_DAMAGE_FACTOR_BATCH.md#remaining-limits-and-next-ten-tasks).
docs/research/ORIGINAL_MODELS_RUST_BATCH.md CHANGED
@@ -49,3 +49,6 @@ Original file buffers are now used directly, but native loader18183 allocation/e
49
  8. Continue shared handler293732 through actual trace/contact/statistics with original-model geometry rather than supplied outcomes.
50
  9. Recover successful295018 pellet target/location aggregation, four-target/fifteen-slot bookkeeping and contact conditions.
51
  10. Build and validate the first interactive macOS/Linux Rust/Bevy/Solari gameplay slice, continuing maps/materials/textures/audio/network/lifecycle work and platform packaging.
 
 
 
 
49
  8. Continue shared handler293732 through actual trace/contact/statistics with original-model geometry rather than supplied outcomes.
50
  9. Recover successful295018 pellet target/location aggregation, four-target/fifteen-slot bookkeeping and contact conditions.
51
  10. Build and validate the first interactive macOS/Linux Rust/Bevy/Solari gameplay slice, continuing maps/materials/textures/audio/network/lifecycle work and platform packaging.
52
+
53
+
54
+ Follow-up: [RE-703–712](QVM_DAMAGE_FACTOR_BATCH.md) completes289594 factor/wound logic on original contexts and the contact dispatch suffix.211239 health damage and complete292545 composition remain open; RE-702 is partially advanced and RE-713 tracks the next work.
docs/research/QVM_CONTACT_DISPATCH.md ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ # Contact direction basis and health-damage dispatch
2
+
3
+ RE-712. Reproduce with `python3 scripts/analyze_qvm_contact_dispatch.py`; check with `python3 scripts/test_qvm_contact_dispatch.py`. The [machine report](qvm-contact-dispatch.json) retains source identity and bounded fixtures.
4
+
5
+ Actual293373–293627 converts attacker pitch/yaw/roll at60/64/68 using word1016003125, invokes six explicitly modeled native sin/cos calls, and stores global forward36477748, right36477736 and up36477724. The complete nine-argument211239 request is target, attacker, attacker, forward pointer, trace point pointer, truncating integer scalar, zero flags, original descriptor4200+432×weapon, location. A nonzero scalar and zero special flag are required; nonzero subunit scalars still dispatch integer0, and explicit negative fixtures dispatch signed negative integers. Signed zero skips dispatch. All paths return the unchanged float scalar. Original nearest-point words supply trace points. Native transcendental parity, earlier contact/event stages and actual211239 health/armor mutations remain open;211239 is captured without effects.
6
+
7
+ No original executable/full QVM/native collision/physics/rendering executes. Native responses and legal input domains are supplied or unverified where stated; full runtime/macOS/Linux Rust/Bevy/Solari gameplay remains open.
docs/research/QVM_DAMAGE_FACTOR_BATCH.md ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Damage factor and contact dispatch: RE-703–712
2
+
3
+ Ten separately checked tasks extend the [original-model/Rust batch](ORIGINAL_MODELS_RUST_BATCH.md). Complete289594 now computes factors and wound bookkeeping from original-model contact contexts. The contact caller's direction basis, damage-call arguments and scalar return are recovered separately. This batch adds no playable Rust/Bevy/Solari behavior.
4
+
5
+ | Task | Completed scope | Evidence |
6
+ | --- | --- | --- |
7
+ | RE-703 | Protection gates and actual armor inventory selection | [Protection](QVM_DAMAGE_PROTECTION.md) |
8
+ | RE-704 | Original weapon/location descriptor pairs and cubic range local | [Range](QVM_DAMAGE_RANGE.md) |
9
+ | RE-705 | Weapon24 half/full-range attenuation | [Attenuation](QVM_DAMAGE_ATTENUATION.md) |
10
+ | RE-706 | Factor saturation, weapon4 cap and optional accumulation | [Clamp](QVM_DAMAGE_CLAMP.md) |
11
+ | RE-707 | Wound eligibility, leg flag and client4312 increments | [Wounds](QVM_DAMAGE_WOUNDS.md) |
12
+ | RE-708 | Attacker/weapon attribution, counter and pending flag | [Attribution](QVM_DAMAGE_ATTRIBUTION.md) |
13
+ | RE-709 | Original location jump table and client192 flags | [Location flags](QVM_DAMAGE_LOCATION_FLAGS.md) |
14
+ | RE-710 | Held-item deadline and final return override | [Interruption](QVM_DAMAGE_INTERRUPT.md) |
15
+ | RE-711 | Complete factor function with original-model contact contexts | [Complete factor](QVM_DAMAGE_COMPLETE.md) |
16
+ | RE-712 | Contact direction basis, captured health-damage ABI and return | [Contact dispatch](QVM_CONTACT_DISPATCH.md) |
17
+
18
+ ## Findings and validation
19
+
20
+ Mode9 and strict spawn-protection gates precede damage location selection. Actual32155/32117 checks positive client health and16 inventory slots, comparing low-byte item IDs17/22 for vest/helmet classification. Source descriptors use4068+432×weapon+8×location for factor/wound pairs. Weapon24 attenuates from scale1 at half range to float32(0.1) at full range. Factors saturate at1, weapon4 caps at float32(0.35), and its optional accumulator adds each capped factor without a total clamp.
21
+
22
+ Weapon4 computes a clamped distance ratio and cubic complement, but complete289594 never consumes the cubic local in its return arithmetic. Its distance ratio still gates wound-meter effects below float32(0.15). Client/entity flags suppress wounds; accepted wounds update a meter, attribution, a counter capped by a strict below4 check, and pending flag256. The original location table independently sets grouped client192 bits. Held items with low bytes14/23 can write time+200 deadlines above factor thresholds0.2/0.5, except leg locations11–14. A final global override returns1 after preceding effects; early zero/protection returns bypass it.
23
+
24
+ The complete factor fixture copies original hit contexts verbatim into a separate memory region and verifies report identity. Its independent oracle checks return, location output, accumulator, tracked external client fields and attacker attribution. Source table words are checked directly against original file bytes. Explicit synthetic clock/range/counter/flag fixtures expose arithmetic boundaries without establishing legal live-game domains.
25
+
26
+ Contact293373–293627 rebuilds forward/right/up from attacker angles using six modeled native sin/cos calls. It captures211239 only when the scalar is nonzero and the special flag is zero. Arguments preserve target, attacker twice, forward pointer, trace point pointer, scalar truncated to integer, zero flags, original weapon method and location. Nonzero subunit scalars dispatch integer0; explicit negative scalars dispatch signed negatives. The caller returns its unchanged float scalar.
27
+
28
+ The1259-test baseline gains41 Python checks, ending at1300 passing tests. All ten new JSON reports regenerate byte-identically, each below2MiB. Storage/filter, documentation-link and whitespace checks pass. Earlier reports, original assets, Rust code and the shared VM evaluator remain unchanged. Original native/executable parity and Linux execution remain open.
29
+
30
+ ## Remaining limits and next ten tasks
31
+
32
+ RE-702 is partially advanced for contact/damage analysis, with its Bevy/native/interactive scope still open. RE-713 tracks the next work.211239 remains captured, so these results establish no health/armor mutations. Full292545 is not composed yet: ordinary/special location selection, contact events and feedback still separate the accepted prefix from the recovered dispatch suffix. Native sin/cos results are modeled, not reference-captured.
33
+
34
+ 1. Compose ordinary contact292545 with complete289594, including scalar×100 and location-zero suppression.
35
+ 2. Recover special weapon4 contact location/equipment selection and its bypass of ordinary health dispatch.
36
+ 3. Recover contact event allocation, normal encoding, IDs, flags and health prediction at292917 onward.
37
+ 4. Recover event scalar/accumulator encoding, direction normalization, weapon27 muzzle handling and feedback gates.
38
+ 5. Compose complete292545, including impact-only and nonclient branches, with actual recovered children and explicit native boundaries.
39
+ 6. Recover211239 health/armor/team/knockback/death gates and mutations against independently checked boundary fixtures.
40
+ 7. Compose shared293732 trace/contact/statistics and successful295018 pellet aggregation using original-model geometry.
41
+ 8. Recover native loader18183 ownership/endian/failure behavior and capture native/executable parity for geometry, transcendental math and live frame/clock domains.
42
+ 9. Add Rust frame/tag interpolation, hierarchy, geometry/classification contracts and Bevy asset resolution, preserving case and archive variants.
43
+ 10. Build an interactive macOS/Linux Bevy/Solari slice and continue maps, materials, textures, audio, networking, lifecycle and packaging validation.
docs/research/README.md CHANGED
@@ -735,3 +735,9 @@ The extension collector adds five focused tests for native tail calls, command-r
735
 
736
 
737
  [RE-711: Complete damage factor on original-model contacts](QVM_DAMAGE_COMPLETE.md). Reproduce with `python3 scripts/analyze_qvm_damage_complete.py`; check with `python3 scripts/test_qvm_damage_complete.py`.
 
 
 
 
 
 
 
735
 
736
 
737
  [RE-711: Complete damage factor on original-model contacts](QVM_DAMAGE_COMPLETE.md). Reproduce with `python3 scripts/analyze_qvm_damage_complete.py`; check with `python3 scripts/test_qvm_damage_complete.py`.
738
+
739
+
740
+ [RE-712: Contact direction basis and health-damage dispatch](QVM_CONTACT_DISPATCH.md). Reproduce with `python3 scripts/analyze_qvm_contact_dispatch.py`; check with `python3 scripts/test_qvm_contact_dispatch.py`.
741
+
742
+
743
+ [RE-703–712 batch overview and next tasks](QVM_DAMAGE_FACTOR_BATCH.md).
docs/research/qvm-contact-dispatch.json ADDED
The diff for this file is too large to render. See raw diff
 
scripts/analyze_qvm_contact_dispatch.py ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python3
2
+ """Contact angle basis, health-damage request ABI and scalar return."""
3
+ import hashlib,itertools,json,math
4
+ from pathlib import Path
5
+ from qvm_research_support import source,constants
6
+ from qvm_trace_support import resume,captured
7
+ from analyze_qvm_contact_gates import LOCAL,ATTACKER,TARGET,TRACE
8
+ from analyze_qvm_acceleration import word,number
9
+ from analyze_qvm import write_json
10
+ FORWARD=36477748
11
+ RIGHT=36477736
12
+ UP=36477724
13
+
14
+ def dispatch(ins,m):
15
+ requests=[]
16
+ def trig(fn):
17
+ def call(i,m,a):
18
+ result=word(fn(number(a[8])));return m,result,{'input_word':a[8]&0xffffffff,'result_word':result,'native_math_modeled':True}
19
+ return call
20
+ m,e=resume(ins,293373,293628,m,LOCAL,handlers={-104:trig(math.sin),-105:trig(math.cos),211239:captured(211239,requests)})
21
+ return m,{'basis':{name:[m[p+4*j]&0xffffffff for j in range(3)] for name,p in [('forward',FORWARD),('right',RIGHT),('up',UP)]},'requests':requests,'execution':e}
22
+
23
+ def main():
24
+ q,raw,h,ins=source();path=Path('docs/research/md3-original-model-hit.json');data=path.read_bytes();original=json.loads(data)['probes'];rows=[]
25
+ for angles,w,scalar,special in itertools.product([(0,0,0),(30,45,60),(-89,270,-15),(180,-720,90)],[3,4,24],[-2.9,-0.0,0,0.9,35.75],[False,True]):
26
+ row=1 if w!=24 else 2;context=original[row]['context'];loc=context[17];a=dict(angles=angles,weapon=w,scalar=scalar,special=special,location=loc,context_row=row);m=constants(raw,h,[4200+432*w]);mod=m[4200+432*w]
27
+ m.update({LOCAL+132:ATTACKER,LOCAL+136:TRACE,LOCAL+140:w,LOCAL+48:TARGET,LOCAL+52:int(special),LOCAL+56:word(scalar),LOCAL+60:loc,TARGET+684:100})
28
+ for j,angle in enumerate(angles):m[ATTACKER+60+4*j]=word(angle)
29
+ for j,v in enumerate(context[3:6]):m[TRACE+12+4*j]=v
30
+ m,p=dispatch(ins,m);p.update(input=a,method_word=mod,trace_point_words=context[3:6],target_health=m[TARGET+684]);rows.append(p)
31
+ write_json(Path('docs/research/qvm-contact-dispatch.json'),{'source':q['source'],'sha256':q['sha256'],'context_report':str(path),'context_report_sha256':hashlib.sha256(data).hexdigest(),'probes':rows,'limits':'Actual293373..293627 with modeled native sin/cos returning float32;211239 captured only. Original-model nearest-point words supply trace point. Earlier contact/event gates are not run. No health damage or original-native transcendental parity.'})
32
+ if __name__=='__main__':main()
scripts/test_qvm_contact_dispatch.py ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import hashlib,json,math,unittest
2
+ from pathlib import Path
3
+ from test_qvm_ground import word,n,f
4
+ from analyze_qvm_contact_gates import ATTACKER,TARGET,TRACE
5
+ from analyze_qvm_contact_dispatch import FORWARD
6
+
7
+ def basis(angles):
8
+ radians=[f(f(a)*n(1016003125)) for a in angles];sp,sy,sr=[f(math.sin(a)) for a in radians];cp,cy,cr=[f(math.cos(a)) for a in radians]
9
+ mul=lambda a,b:f(a*b)
10
+ return {'forward':list(map(word,[mul(cp,cy),mul(cp,sy),-sp])),
11
+ 'right':list(map(word,[f(mul(mul(-sr,sp),cy)+mul(-cr,-sy)),f(mul(mul(-sr,sp),sy)+mul(-cr,cy)),mul(-sr,cp)])),
12
+ 'up':list(map(word,[f(mul(mul(cr,sp),cy)+mul(-sr,-sy)),f(mul(mul(cr,sp),sy)+mul(-sr,cy)),mul(cr,cp)]))}
13
+
14
+ class ContactDispatchTests(unittest.TestCase):
15
+ @classmethod
16
+ def setUpClass(cls):cls.r=json.loads(Path('docs/research/qvm-contact-dispatch.json').read_text());cls.ps=cls.r['probes']
17
+ def test_independent_source_order_basis_words(self):
18
+ for p in self.ps:self.assertEqual(p['basis'],basis(p['input']['angles']))
19
+ def test_exact_damage_call_arguments_and_truncation(self):
20
+ for p in self.ps:
21
+ a=p['input'];called=f(a['scalar'])!=0 and not a['special'];self.assertEqual(len(p['requests']),int(called))
22
+ if called:self.assertEqual(p['requests'][0]['arguments'],{'8':TARGET,'12':ATTACKER,'16':ATTACKER,'20':FORWARD,'24':TRACE+12,'28':math.trunc(f(a['scalar'])),'32':0,'36':p['method_word'],'40':a['location']})
23
+ def test_return_unchanged_scalar_and_health_not_executed(self):
24
+ self.assertEqual(len(self.ps),120)
25
+ for p in self.ps:self.assertEqual(p['execution']['return']&0xffffffff,word(p['input']['scalar']));self.assertEqual(p['target_health'],100);self.assertEqual(len([c for c in p['execution']['calls'] if c['target']<0]),6)
26
+ def test_original_point_provenance(self):
27
+ data=Path(self.r['context_report']).read_bytes();self.assertEqual(hashlib.sha256(data).hexdigest(),self.r['context_report_sha256']);rows=json.loads(data)['probes']
28
+ for p in self.ps:self.assertEqual(p['trace_point_words'],rows[p['input']['context_row']]['context'][3:6])
29
+ if __name__=='__main__':unittest.main()