[napetrov]

#153 took _provenance_warnings to complexity 11, over the cap of 10, so test_growth_ratchet fails on main and on every PR's merge commit (seen first on #186).

This moves the resolved-model and uptake checks into _arm_warnings. The warnings and their order don't change.

Ran the full suite locally: 7245 passed. ruff and mypy are clean on compare.py.

🤖 Generated with Claude Code

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.24%. Comparing base (af17696) to head (df2fd99).
⚠️ Report is 1 commits behind head on main.
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #195   +/-   ##
=======================================
  Coverage   92.24%   92.24%           
=======================================
  Files         239      239           
  Lines       29934    29938    +4     
  Branches     4917     4917           
=======================================
+ Hits        27613    27617    +4     
  Misses       1633     1633           
  Partials      688      688           

Files with missing lines	Coverage Δ
dokimi/dokimi/repos/compare.py	99.68% <100.00%> (+<0.01%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[antonwolfy]

LGTM. The split is a verbatim move, and it brings _provenance_warnings from complexity 11 to 8, with _arm_warnings at 4, so the ratchet passes again. No findings.

What I ran, on a Linux review host at ce3daad:

• make ci-local: 27 passed, 2 failed, 7 skipped; 7246 tests passed. The changelog FAIL is covered by the skip-changelog label. The pip-audit FAIL reports a gitpython 3.2.0 advisory (fix in 3.2.1); this PR touches no dependency files.
• scripts/check_growth.py and tests/test_growth_ratchet.py: both pass (46 tests). compare.py is 503 lines, under the cap.
• A mutation run on _arm_warnings: 9 mutants against the compare, uptake, host and run tests, 8 killed (dropped uptake check, dropped models filter, inverted used, swapped arms, models=None at the call site, and others).

Checked, no issues

• Order: the call sits at the old spot in the per-treatment loop, after the stamp and harness-version warnings, so the output order is unchanged. The trailing mismatch_warnings still comes last.
• Inputs: cells, baseline, text and models are everything the moved block used; both warning strings are identical.
• References: nothing else in the repo names _provenance_warnings or _arm_warnings, so no docs need updating.

[antonwolfy]

@napetrov merged into the current main (565f653), this PR fails these tests:

• tests/test_growth_ratchet.py::test_no_file_or_function_grew_past_its_baseline
• tests/test_growth_ratchet.py::test_no_file_or_function_grew_past_its_baseline

The PR's checks ran before main moved to 565f653, so they did not see it. Merging main into the branch reproduces it; once that is fixed I can merge.

Dismissing until the growth-baseline entry for _provenance_warnings is dropped (see the comment above: merged into the current main, test_growth_ratchet fails). I'll review again after that.

[napetrov]

Merged intel/main and fixed the growth ratchet failure in a16a5d4: main carried a baseline entry for compare.py::_provenance_warnings (complexity 11); with the split it is 8, so the entry is removed (scripts/check_growth.py --update). tests/test_growth_ratchet.py passes at this head.

[antonwolfy]

LGTM. The split is still a verbatim move, and the merges of main and the removed _provenance_warnings baseline entry are correct. No findings.

What I ran, on a Linux review host at df2fd99:

• make ci-local: 28 passed, 1 failed, 7 skipped; 8,195 tests passed; the changed-lines coverage check passes. The changelog FAIL is covered by the skip-changelog label.
• scripts/check_growth.py and tests/test_growth_ratchet.py: both pass (46 tests). _provenance_warnings is at complexity 8 and _arm_warnings at 4, under the cap of 10.
• A mutation run on _arm_warnings and its call site: 15 mutants against the compare, uptake and host tests; the dropped uptake check, dropped models filter, swapped arms, models=None at the call site and the other behavioural mutants were all killed.

Checked, no issues

• Equivalence: the body of _arm_warnings matches the removed inline block character for character; the call sits at the old spot, so the warning order is unchanged and mismatch_warnings still comes last.
• Baseline: the removed growth_baseline.json entry is the only one that had to go, and the file stays valid JSON.
• References: nothing else names _provenance_warnings or _arm_warnings, so no docs need updating.