Files
timothy 7be42654fe
PR Gates / CI image pin matches docker/ci (pull_request) Successful in 11s
PR Gates / Docs update reminder (pull_request) Successful in 18s
Review verdict / Set review-verdict status (pull_request_target) Successful in 8s
Build ErsatzTV Image / API docs in sync (OpenAPI + endpoint index) (pull_request) Successful in 1m33s
Build ErsatzTV Image / Formatting (changed .cs conform to .editorconfig) (pull_request) Successful in 1m21s
review-verdict/h10 Review-verdict: MERGEABLE @ 7be4265 (base: main)
PR Gates / decisions lifecycle (pull_request) Successful in 2m7s
PR Gates / Script tests (pytest) (pull_request) Successful in 2m13s
Build ErsatzTV Image / Build & test (.NET) (pull_request) Successful in 9m7s
Build ErsatzTV Image / Functional E2E (curl + UI contracts) (pull_request) Successful in 17m2s
Build ErsatzTV Image / EF migration integrity (SQLite + MySql) (pull_request) Successful in 18m58s
Build ErsatzTV Image / Build & push image (amd64) (pull_request) Has been skipped
fix(685): suppress both empty-states on error; name the addable-kind derivation
Round-4 verification returned MERGEABLE with no new defects and no
BLOCKER/HIGH/MEDIUM. These are its remaining LOW and nits.

- "No results — try a search above." asserts a search that COMPLETED and found
  nothing, so it was false beside a failed request. Both empty-state messages
  are now suppressed on error and the role="alert" banner is the whole message,
  as that test's comment already claimed. Pinned positively and negatively so a
  refactor cannot satisfy the assertion by rendering nothing at all.

- The (typeof ADDABLE_TYPE_LIST)[number] derivation is now the named
  AddableKind, spelled once instead of twice: a third ingress into the searched
  kinds is most likely to be written by copying one of the existing two, and the
  "one list, all ingresses" property should be visible at a glance rather than
  reassembled.

- The §3b lesson cited two measured test counts, which go stale against the very
  suite they describe — a count taken before the helper had unit tests no longer
  holds now that it does. Scoped the observation to the sha it was measured on
  and replaced the counts with the invariant they were evidence for: every gate
  needs at least one test that reddens when that gate ALONE is removed.

- Documented why the error banner stays conditionally mounted while the hint's
  live region does not: role="alert" is the one live-region role screen readers
  reliably announce on insertion, so the two are correct for opposite reasons.
  The review flagged the divergence as unexplained, not as wrong.

refs #740
2026-08-05 19:40:07 +02:00
..
2026-07-07 10:05:39 +02:00
2026-07-02 18:36:56 +02:00
2026-07-07 10:05:39 +02:00