fix: repair settings and racy labeling test found during analysis - #3374
fix: repair settings and racy labeling test found during analysis#3374Vladyslavleiner wants to merge 1 commit into
Conversation
![graphify-labs[bot]](https://avatars.githubusercontent.com/in/4356270?s=60&v=4){kind=small}
There was a problem hiding this comment.
Graphify reviewed this change.
Looks safe to merge — no coupling regressions and no blocking issues, checked against the code graph (not a self-assessment).
Graphify review — findings
Bumps the security policy's supported versions from 0.3.x to 0.9.x and the project version to 0.9.55. Pins test_label_communities_batches_when_over_batch_size to max_concurrency=1 so batch calls record in deterministic order, with the concurrent ordering covered separately. Regenerates uv.lock with reworked dependency markers.
No blocking issues surfaced. 3 lower-confidence candidates did not survive cross-model review.
Analysis details — impact, health, verification
Impact & health
Graphify review
Impact — 84 functions depend on the 84 functions this change touches.
Health — grade A; no new coupling hotspots.
Verification — 84 functions in the blast radius were not formally verified this run (proofs are advisory here).
Gate & verification
graphify gate
PASS — objectively clean (no health regressions, tests not run — proofs not run this pass (advisory)). Grounded, not self-assessed.
Advisory (not blocking):
- verification_scope: 84 function(s) in the blast radius were not formally verified this run
- tests/test_labeling.py: pin max_concurrency=1 in the batch-order test so the recorded batch sizes are deterministic (default concurrent path can complete batches out of order) - SECURITY.md: bump supported versions to 0.9.x (table was stale at 0.3.x) - uv.lock: regenerate to match graphifyy 0.9.55 (was pinned at 0.9.53)
…erministic The assertion recorded the order batches reached their LLM call inside worker threads, which is nondeterministic; the product merges by cid on the main thread and concurrency equivalence is covered by a sibling test. Takes only the flaky-test fix from #3374 (its SECURITY.md/uv.lock changes are handled separately).
1 participant
Summary
Fixes found while running the full test suite and auditing settings.
Changes
tests/test_labeling.py— pinmax_concurrency=1in the batch-order test so the recorded batch sizes are deterministic. The default concurrent path (Feature request:--max-concurrencyand--batch-sizeforcluster-only/label#1390) can complete batches out of order, makingassert calls == [100, 100, 50]flaky (observed[100, 50, 100]on Python 3.14).SECURITY.md— bump the supported-versions table from0.3.xto0.9.x(the project is at 0.9.55).uv.lock— regenerate so the pinnedgraphifyyversion matches pyproject (0.9.53->0.9.55).Verification
uv lock --checkpassesuv sync --frozenpasses5347 passed, 88 skipped(aftergit fetch --unshallowand installing extras)