This release fixes issues for SREs watching stability and regressions.
✓ No known CVEs patched in this version
Topics
+12 more
Summary
AI summaryLowered TEST_PAIRING_MIN_SAMPLE from 10 to 5 to fix stale test advisories.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Bugfix | Medium |
Lowered TEST_PAIRING_MIN_SAMPLE from 10 to 5 to match MIN_SAMPLE_SIZE. Lowered TEST_PAIRING_MIN_SAMPLE from 10 to 5 to match MIN_SAMPLE_SIZE. Source: llm_adapter@2026-07-20 Confidence: medium |
— |
Full changelog
Fixed
-
TEST_PAIRING_MIN_SAMPLEwas left at 10 when the sibling convention floors
were measured down to 5 (v4.4.23), so test-pairing derived for almost nothing
even after v4.4.43 taught it the flattests/layout. The two thresholds
answer the same question -- how many siblings make a convention trustworthy --
andconventions.pysays outright that they "should not disagree". 10 sits
ABOVE the natural cohort size of a real repo: across the fixture matrix only
1 of 7 archetypes (ts-plain), 1 of 12 (py-django) and 1 of 10 (rb-plain) clear
it, so a fully-paired repo still producedtest_pairing: {}and the stale-test
advisory riding on it stayed silent. Lowered to 5 to matchMIN_SAMPLE_SIZE.test_below_sample_floor_returns_emptywas sized at a literal 9 (chosen when
the floor was 10) and would have silently stopped testing anything; it now
sizes offMIN_SAMPLE_SIZEitself so the floor and its guard cannot drift
apart again.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About Chameleon
All releases →Beta — feedback welcome: [email protected]