Keep original-source selection usable through long histories and refreshes - #2870
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Exact-head validation snapshot and review entry pointHead: The hosted Smart CI Self-Test passed on configured Node 24.13.1: 506 tests passed, 0 failed/skipped/cancelled, and the action-pin inventory passed 161/161 immutable references. Run: https://github.com/Chris0Jeky/Taskdeck/actions/runs/34423790161 ; job: https://github.com/Chris0Jeky/Taskdeck/actions/runs/34423790161/job/102704612600 . The actual job log was read, not inferred from local fixtures.
Local evidence remains separate: 357 final overlay/placement regressions and an independent 309-test portable export run, zero failures/skips/cancellations. Earlier hosted and local fixture failures/corrections remain documented; no retry erased their record. Review entry point: https://github.com/Chris0Jeky/Taskdeck/blob/3508ef0d3b8f1b0eca34125c18007efa756ab3b0/docs/ci/continuation/OPERATIONS.md All seven current heads have passed hosted Smart CI Self-Test and CI Extended. All seven PRs remain OPEN/unmerged and require independent + maintainer review. This is a snapshot, not a promise to monitor later results. Production Taskdeck result reuse, canonical admission integration and protected execution-provenance infrastructure remain explicitly unactivated/outstanding. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: ac13824640
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| <div class="original-choices"> | ||
| <button v-if="!loaded || nextOffset !== null" type="button" :disabled="disabled || loading" @click="load"> | ||
| <button v-if="!loaded || nextOffset !== null" type="button" :disabled="disabled || loading" | ||
| :aria-label="`${loading ? 'Checking originals' : loaded ? 'Load more originals' : 'Choose original sources'} for ${memoryTitle || 'this memory'}`" @click="load"> |
There was a problem hiding this comment.
When a board contains two memories with the same title—which the memory model permits and generated memories can produce—both controls receive the same accessible name (for example, Choose original sources for Thinking question). Screen-reader users and role-based automation therefore still cannot identify which memory each button controls, despite this change's goal of making the pickers distinguishable. Include an additional unique visible/accessibility discriminator such as the memory version or an associated memory label.
Useful? React with 👍 / 👎.
b262c8e
into
codex/2808-overhaul-integration
1 participant
Private original histories stopped at an arbitrary offset, indistinguishable source buttons were difficult to use, and refresh/access transitions could discard selections or retain stale content. This continuation of #2808 removes the history cutoff with bounded pages, names original pickers by memory, retracts revoked source contents, preserves same-user token refresh, and disables retained-instruction sends while thinking or receipt recovery blocks the composer.
This PR is stacked on #2866. It adds no board mutation or external model call.
Verification: full backend solution passed 9,181 tests with 34 existing skips; full frontend passed 6,358 with three existing skips. Targeted API 27 and frontend 33, typecheck, production build, scoped ESLint, two Chromium journeys, doc links and governance passed. The SQLite regression retains 1,012 immutable originals and follows pages beyond the old cutoff. Browser evidence covers distinguishable memory choices, revoked access and recovery, four experiences and 375 px. Working database untouched.
Independent Luna review found no HIGH/CRITICAL defect. MEDIUM follow-through: deep SQL offset pagination can incur increasing scan work; keyset pagination is tracked under #2808 source scalability. The UI keeps ten-item pages. Dedicated 404 and parent retained-send integration cases are nonblocking coverage follow-through. Hosted checks remain pending on this exact head. OUTSTANDING_TASKS.md owner decisions are unchanged.
Parent audio repair is integrated at ac13824. No source-interaction logic changed during integration. Combined source/header API35 and frontend38 passed, along with docs. The parent repair also has its own production-policy/browser evidence and scoped independent review; hosted checks must qualify this merged head.