[aozmekik]

SUMMARY

babel-extract takes its baseline from the tip of the base branch when the job starts, while the checkout is the merge ref GitHub built when the event fired. If something lands on master in between, the job counts that change against the PR. After #45741 de-fuzzed 196 Turkish entries, the queued runs for #45709 and #45640 and the master push run for ff0af9c failed with tr: 196 confirmed translation(s) invalidated, although none of them touch translations.

On a direct push the fetch gets the branch tip too, so the pushed commit is usually compared with itself rather than with the previous commit.

This uses the first parent of $GITHUB_SHA as the baseline: the base commit of the merge ref for a PR, the previous commit for a push.

TESTING INSTRUCTIONS

I ran the job's steps (babel_update.sh on both sides, then check_translation_regression.py) on 6cfbbe2, the #45709 merge ref from the failing run, and on ff0af9c:

• baseline f93aa5b (master tip when those jobs ran): tr: 196 confirmed translation(s) invalidated, same as CI
• baseline $GITHUB_SHA^1 (ff0af9c and 3d1273b): No translation regressions., tr: translated 4046 -> 4046 (+0), fuzzy 795 -> 795 (+0)

ADDITIONAL INFORMATION

• Has associated issue:
• Required feature flags:
• Changes UI
• Includes DB Migration (follow approval process in SIP-59)
  • Migration is atomic, supports rollback & is backwards-compatible
  • Confirm DB migration upgrade and downgrade tested
  • Runtime estimates and downtime expectations provided
• Introduces new feature or API
• Removes existing feature or API

[codeant-ai-for-open-source]

  CodeAnt PR Risk

🟢  Low Risk

• The PR appears safe to merge; PR checks compare against the merge commit’s first parent, and master pushes are squash commits already checked as PRs.

Assessed commit: eed009076c7d

[hainenber]

It's a LGTM. Thanks, once more.

[sadpandajoe]

lgtm