Skip to content

test: isolate text stats validation - #1289

Merged
sourishkrout merged 3 commits into
mainfrom
codex/text-stats-python-validation
Aug 7, 2026
Merged

test: isolate text stats validation#1289
sourishkrout merged 3 commits into
mainfrom
codex/text-stats-python-validation

Conversation

@sourishkrout

Copy link
Copy Markdown
Contributor

Summary

  • replace repository-wide runme run lint with isolated Python compilation
  • make the oracle solution and workflow criterion use the same validation command
  • update trajectory tests for compile validation

Why

The eval's validation command unintentionally resolved the parent repository's lint workflow locally and was unavailable in Docker. That coupled an isolated Python task to Runme's Go repository and toolchain.

Impact

Validation now behaves consistently across host-backed and Docker environments. It no longer depends on parent repository instructions or on Runme being installed in the eval container.

Validation

  • uv run --project integrations/harbor pytest -q integrations/harbor/tests/test_text_stats_reward_criteria.py (32 passed)
  • Docker environment build and oracle solution execution
  • TZ=UTC runme run check test

@sourishkrout
sourishkrout marked this pull request as ready for review August 7, 2026 16:25
Signed-off-by: Sebastian (Tiedtke) Huckleberry <sebastiantiedtke@gmail.com>
Signed-off-by: Sebastian (Tiedtke) Huckleberry <sebastiantiedtke@gmail.com>
@sourishkrout
sourishkrout force-pushed the codex/text-stats-python-validation branch from 916db3f to 18ef22d Compare August 7, 2026 16:26
Signed-off-by: Sebastian (Tiedtke) Huckleberry <sebastiantiedtke@gmail.com>
@sourishkrout
sourishkrout merged commit 9f55ed1 into main Aug 7, 2026
7 checks passed
@sourishkrout
sourishkrout deleted the codex/text-stats-python-validation branch August 7, 2026 16:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant