Record FastLanes Linux retry evidence - #172
Conversation
![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: 52abb032f3
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".
| "artifact_name": "data-format-lab-fastlanes-linux-x86_64-29520329351", | ||
| "results_path": "runs/fastlanes-linux-29520329351/results.json", | ||
| "archive_path": "outputs/release/data-format-lab-fastlanes-linux-x86_64-29520329351.tar.zst", | ||
| "archive_sha256": "b9270a10d9cc0f5d81a81e67b8e22af591a06018e767720b9d9c6e4c3689227a" |
There was a problem hiding this comment.
The checked .github/workflows/benchmark-fastlanes.yml uploads this archive only as a GitHub Actions artifact with retention-days: 14, while this record provides no durable Release asset URL. After that retention window, results.json, logs, case inputs, and the archive named here will be unavailable, leaving only an unverifiable summary and hash; attach the archive to a Release as required by docs/native-robustness.md before publishing this attempt.
Useful? React with 👍 / 👎.
| "package": "pyfastlanes==0.1.3.post9", | ||
| "runtime_version": "0.1.4" | ||
| }, | ||
| "build_command": "python -m venv native/fastlanes/.venv; pip install scikit-build-core==0.11.6 cmake==4.4.0 ninja==1.13.0 pybind11==2.12.1 setuptools-scm==8.3.1; CMAKE_BUILD_PARALLEL_LEVEL=4 pip install --no-build-isolation -e native/fastlanes", |
There was a problem hiding this comment.
When this command is run literally from a clean checkout, it creates native/fastlanes/.venv but then invokes unqualified pip for both installs, so they target the caller's ambient interpreter rather than the new virtualenv. The workflow actually uses native/fastlanes/.venv/bin/python -m pip inside nix develop; recording that executable-qualified invocation is necessary to reproduce the tested build without mixing interpreters and toolchains.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4f79a3c09e
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".
| "cmake": "4.1.2", | ||
| "ninja": "1.13.2", |
There was a problem hiding this comment.
For this Linux attempt, these values conflict with the recorded build command: the workflow installs CMake 4.4.0 and Ninja 1.13.0 into the FastLanes virtualenv before invoking scikit-build-core, while its metadata step prints the separate Nix-shell cmake and ninja versions, 4.1.2 and 1.13.2. The research log consequently describes the build as using the latter versions without establishing which executables performed the build. Distinguish the ambient and build-tool versions and record the venv tool output so reproductions do not silently use a different toolchain.
Useful? React with 👍 / 👎.
1 participant
Summary
The result stays PARTIAL / FAILED and is excluded from all rankings.
Refs #38
Verification