Repository navigation
Conversation
|
Superseded by #172, which records the same Linux run with the complete per-case results, environment, archive SHA-256, regression assertions, and the updated selection guide. Closing this duplicate to keep the non-stacked history clear. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c901196a52
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| "package": "pyfastlanes==0.1.3.post9", | ||
| "runtime": "0.1.4" | ||
| }, | ||
| "result": "numeric and all five string cases exited SIGSEGV; the mixed 13-column case raised RuntimeError UNREACHABLE; malformed comma input raised RuntimeError. The workflow preserved per-case inputs, artifacts, logs, and SHA-256 hashes." |
There was a problem hiding this comment.
The only referenced copies of the per-case files and SHA-256 values are uploaded by .github/workflows/benchmark-fastlanes.yml:292-301 with retention-days: 14; none of the hashes or an archive digest is included in this record. After that window, the workflow link can no longer substantiate this newly recorded result, so persist the hashes here or publish the archive at a durable location.
Useful? React with 👍 / 👎.
| "package": "pyfastlanes==0.1.3.post9", | ||
| "runtime": "0.1.4" | ||
| }, | ||
| "result": "numeric and all five string cases exited SIGSEGV; the mixed 13-column case raised RuntimeError UNREACHABLE; malformed comma input raised RuntimeError. The workflow preserved per-case inputs, artifacts, logs, and SHA-256 hashes." |
There was a problem hiding this comment.
This attempt newly reports failures for both the numeric control and mixed-schema case, but retry_when at line 92 still requires only a successful build, partial-vector strings, and malformed-input handling. A retry could therefore satisfy every documented condition while still reproducing these two fatal failures; extend the criteria to require the numeric and mixed round trips as well.
Useful? React with 👍 / 👎.
| "cmake": "4.1.2", | ||
| "ninja": "1.13.2", |
There was a problem hiding this comment.
The inspected Linux workflow installs cmake==4.4.0 and ninja==1.13.0 into the FastLanes virtual environment immediately before the editable build (.github/workflows/benchmark-fastlanes.yml:74-83), while this evidence lists only the Nix-shell versions 4.1.2 and 1.13.2. This leaves the recorded build environment inconsistent with the workflow; distinguish the shell versions from the virtual-environment build dependencies and record both.
Useful? React with 👍 / 👎.
1 participant
Summary
Related: #38
Verification
python3 -m json.tool research/formats/fastlanes.jsonnix develop --command env PYTHONPATH=src .venv/bin/pytest -q tests/test_research.py tests/test_format_selection.py