cesnimda
22 lines
4.0 KiB
Markdown
22 lines
4.0 KiB
Markdown
# Documents and error navigation
|
||
|
||
Final1440px missing-CV and1024px Documents images were actually re-inspected after the toolbar correction. The final320px Bokmål unknown-route image is byte-identical to the already inspected image. Final source/hash and scoped hygiene evidence are retained; physical/assistive acceptance remains pending.
|
||
|
||
**Implemented CONDITIONAL GO; final hygiene/publication pending.** Existing Documents metadata projection/query/download and CV entry/history owners remain.
|
||
|
||
An ordinary browser regression reproduced a single English action word spanning two lines at1024px. Natural table sizing and word-boundary action wrapping retain complete labels without forcing app overflow. Existing metadata wrapping, owner scoping, sort/filter/page/deep-link and immutable archive behaviors remain.
|
||
|
||
Private CV opening errors now use shared page/read-state presentation: contextual heading, localized missing/denied context, Back and GET-only Retry for temporary failures. Missing/foreign404 and denied401/403 do not offer retry or expose a private ID. Successful retry focuses the existing editing region without a save; prior-owner lifetime cancellation and renderer/pagination remain unchanged.
|
||
|
||
CV history translates only its known source enums, preserving version numbers, restore actions and unknown values. An additional unknown `constructor` regression caught inherited-object lookup; own-key lookup preserves that value verbatim. Current-version metadata uses valid block markup around the existing badge. No revision or content geometry changes.
|
||
|
||
Unknown routes retain the wildcard, Home and browser Back actions, shared surface tokens and a contextual shell h1/owner h2. Two bilingual unit cases prove Back retains its previous query. A first test harness attempted data-router navigation without jsdom Request; the corrected MemoryRouter exercises the same navigation without a new runtime dependency. Both results remain retained.
|
||
|
||
Initial CV entry44-case vector41passed/3failed; error correction then passed44cases. History added one ordinary failing case and its prototype-name extension also failed before own-key lookup. Affected final CV/NotFound/Documents vector **62/62 across three suites PASS** before that last own-key assertion; its targeted rerun passes and the fresh full run covers the final source. Native **10/10 PASS** includes actual deleted-source404, controlled503 recovery with unchanged version/no save, bilingual unknown-route Back, all seven metadata resource kinds and eight responsive/theme samples. These finite results are not human/device/PDF acceptance. Final source/visual/static/publication evidence follows.
|
||
|
||
Raw incomplete triage found the same roleless-label pattern on the existing CV Undo/Redo tooltip wrappers. An assertion in the existing edit/undo/redo regression failed before applying the established describeChild pattern; the complete selected undo/redo workflow passes afterward. Both actual button names, disabled states and data actions remain. No new case or disabled test is introduced.
|
||
|
||
Final native **10/10 PASS**, no skips/flaky/retries,18:21:21–18:22:22UTC;20images/18all-enabled axe scans,0violations/105contrast incomplete observations. The earlier two roleless-toolbar incompletes are absent after the correction. [Portable results](evidence/errors-results.json) preserve final source/result/image hashes, while the prior candidate is archived locally. Three representative images were inspected before the toolbar-only correction:320px Bokmål unknown route,1024px English dark Documents and1440px missing CV; final images will be spot-checked separately. Fixed chrome/full-page and overlap/short-badge contrast remain qualified evidence rather than viewport or human acceptance.
|
||
|
||
Existing data/backend/PDF/AI safeguards and all57 preservation dispositions remain. Shared modal changes receive their separate all-consumer regression; no new dialog infrastructure or Phase8 work is introduced. Fresh full final-source regression belongs to P7-12. Static checks and normal push follow.
|