chore: generalize eval-results LFS rule to catch versioned dumps#96
Merged
Conversation
The two LFS patterns matched only 'eval_results.jsonl' and the now-dead
rig-tagged 'eval_results_rig*.jsonl' naming. The current scheme is
version-tagged ('eval_results_v0.7.0.jsonl'), which neither pattern
matched -- git check-attr reported 'unspecified' for those files, so the
next release dump (eval_results_v0.7.3.jsonl) would have committed as a
plain ~67MB git blob instead of an LFS pointer.
Collapse both into one glob 'eval_results*.jsonl' covering bare,
rig-tagged, and version-tagged variants. Existing v-file pointers are
unchanged; this only closes the gap for future dumps.
antoinezambelli
added a commit
that referenced
this pull request
Jun 1, 2026
#98) The wheel target is scoped to src/forge, but the sdist target had no configuration, so hatchling swept the entire working tree -- pulling the LFS eval datasets (~112MB) and the eval dashboard's node_modules (~97MB, including Windows .exe/.node binaries) into the source distribution. Add a scoped [tool.hatch.build.targets.sdist] excluding both. The sdist drops from ~26MB to ~690KB; src/forge, tests, docs, the dashboard source, and the prebuilt results HTML are all retained. report.py rebuilds the dashboard via npm on demand, so the committed node_modules was never load-bearing; the eval datasets remain in the repo via LFS. Also refresh a stale .gitignore comment to the versioned eval-results naming (post #96).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Generalize the eval-results Git LFS rule so every result-dump naming variant is captured.
Why
The two old patterns matched only the bare
eval_results.jsonland the now-retired rig-tagged naming (eval_results_rig*.jsonl). The current scheme is version-tagged —eval_results_v0.6.0.jsonl,eval_results_v0.7.0.jsonl— which neither pattern matched:The two existing v-files are already in LFS (added when a matching
git lfs trackpattern still existed), but the rule protecting them is gone. So the next release dump (eval_results_v0.7.3.jsonl) would have committed as a plain ~67MB git blob baked permanently into history — the "accidental regular-git churn" this cleans up.The fix
One glob —
eval_results*.jsonl— subsumes bare, rig-tagged, and version-tagged forms, so no future variant slips through.Verification
git check-attr filternow returnslfsforeval_results_v0.7.0.jsonland the futureeval_results_v0.7.3.jsonl.git lfs fsck --pointers= OK..gitattributesonly — no code, no proxy overlap.