Remove legacy site-psk-invalidation topic after one release; case-variant overrides silently dropped by the flattener (verify) #105
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Follow-ups from #100 item 3 / #103:
site-psk-invalidationcompatibility publish/subscribe once every central node runs the build with the unifiedcentral-cache-invalidationtopic (one release).Item 2 checked on main
a2c3a2d2: the suspicion doesn't hold — HandleSetInstanceOverrides (bulk attribute) and the single + bulk native-alarm-source override paths already match exactly and reject a case-variant name, storing nothing (the single attribute path is InstanceService.SetAttributeOverrideAsync, also exact). Improved the error: new CaseMismatchHint adds '(names are case-sensitive; did you mean 'X'?)' when exactly one member matches case-insensitively (none when ambiguous). 3 tests (bulk attribute, single + bulk native) assert rejection + hint + nothing written; they fail without the hint. ManagementService 755/755; slnx 0 warnings. Items 1 (remove legacy site-psk-invalidation after one release) and 3 (direct-SQL edits converge by TTL — documented) remain.Status 2026-11-10: case-variant overrides are rejected with a 'names are case-sensitive; did you mean X?' hint (
a2c3a2d2), never silently dropped — consistent with exact-match naming elsewhere. Remaining: remove the legacysite-psk-invalidationtopic — the replacementcentral-cache-invalidation(153609ef, compat3a31dd2d) is not deployed anywhere yet (last deploy records 2026-10-22). Remove it one release after a DEV/QA deploy that carries the replacement.