CultCache: a unique-index refusal names the least violation; shared refusals vector #99

Merged
libby merged 4 commits from hands/variants-pair-order-cs into main 2026-11-11 10:05:32 +00:00

2026-11-11

Libby
unique-index refusals vector: rows 20-22 load rows 14, 15, 17; README gives the real schema ids
All checks were successful
CultMesh QUIC native / Build win32-x64 (pull_request) Has been skipped
CultMesh QUIC native / Build linux-x64 (pull_request) Has been skipped
CultMesh QUIC native / Release native binaries (pull_request) Has been skipped
CultMesh portability and performance / ubuntu-latest reactive semantics and leases (pull_request) Successful in 29m21s
CultMesh QUIC native / Verify committed binaries match SHA256SUMS (pull_request) Successful in 37s
Rows 20-22 open stores holding the records of rows 14, 15 and 17, so a
runtime with no batch commit (TS) still walks cross-group key order. Over
the TS-walkable rows (5, 8-13, 19-22), hand mutants of CompareRefusals'
first-key compare fail: code point row 22; OrdinalIgnoreCase 20, 22;
InvariantCulture 20, 21; InvariantCultureIgnoreCase 20, 21. Rows 1-19 are
byte-identical. The README now names the sha256 schema ids the store
catalogs carry, not the versions, and counts tests.uq_alias's unique Alt.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-11 09:56:06 +00:00
Libby
cultcache C# tests: pin that a refusal orders keys by code unit, not case
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-11 09:03:39 +00:00