File size: 2,580 Bytes
d0faac4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
# LoLBench Harbor Task: TypeScript_3

This is a Harbor-format LoLBench task generated from `TypeScript_PR-58640_Arbitrary-Module-Namespace-Identifiers`.

Selected source instance:

- Harbor ID: `TypeScript_3`
- PR: `microsoft/TypeScript#58640`
- Base commit: `112e860602b2b46f7a1f179a8f6357f6e4db2370`
- Test modes: `orig, aug, union`
- Default/canonical mode: `union`
- Harbor image: `smartdub26/lolbench-linux:TypeScript_3-1.0.0`
- Docker Hub image: `smartdub26/lolbench-linux:TypeScript_3-1.0.0`

## Image Setup

Prefer the published Docker Hub image when it is available. Harbor refers to
the image as `smartdub26/lolbench-linux:TypeScript_3-1.0.0`, so Docker can pull it
directly when it is not present locally:

```bash
docker pull smartdub26/lolbench-linux:TypeScript_3-1.0.0
```

Only if the Docker Hub image is unavailable, build the single task image
locally:

```bash
docker build -t lolbench/typescript_3:1.0.0 harbor_tasks/TypeScript_3/environment
docker tag lolbench/typescript_3:1.0.0 smartdub26/lolbench-linux:TypeScript_3-1.0.0
```

The image contains the processed source tree at the base commit under
`/workspace/typescript`. Harbor must not mount a host source checkout
for the agent.

## Run an Agent

```bash
.harbor-venv/bin/harbor run \
  -p harbor_tasks/TypeScript_3 \
  -a codex \
  -m "<model>" \
  --job-name TypeScript_3_agent_union \
  --jobs-dir harbor_runs/TypeScript_3 \
  --no-delete \
  --n-concurrent 1 \
  --ve LOLBENCH_SUITE=union
```

After implementing the requirement, the agent must clean any test-file edits
and run:

```bash
lolbench-submit
```

## Evaluate a Patch

Choose one mode with `LOLBENCH_SUITE=orig`, `aug`, or `union`; if unset, the
task defaults to `union`.

```bash
PATCH="$(pwd)/path/to/solution.patch"

.harbor-venv/bin/harbor run \
  -p harbor_tasks/TypeScript_3 \
  -a oracle \
  --job-name TypeScript_3_eval_union \
  --jobs-dir harbor_runs/TypeScript_3 \
  --no-delete \
  --n-concurrent 1 \
  --mounts '[{"type":"bind","source":"'"$PATCH"'","target":"/candidate/solution.patch","read_only":true}]' \
  --ae LOLBENCH_ORACLE_PATCH=/candidate/solution.patch \
  --ve LOLBENCH_SUITE=union
```

Outputs:

```bash
find harbor_runs/TypeScript_3/TypeScript_3_eval_union -path '*/verifier/agent_report.json' -print
find harbor_runs/TypeScript_3/TypeScript_3_eval_union -path '*/verifier/reward.json' -print
```

`tests/docker-compose.yaml` is a verifier-only Harbor overlay. It mounts
`tests/`, `tests/private/`, and `solution/` into the separate verifier
container; none of those paths are mounted into the agent container.