Repository navigation
fix: backport 0.12.1 batch 59 (25 wide-bar fixes) - #32856
Conversation
The pull-request review panel and the artifact comments sidebar each show or hide resolved threads with a button that had no aria-expanded, so a screen reader could not tell whether resolved threads were showing. The sidebar's check icon was also exposed to assistive tech. Both buttons now carry aria-expanded, and the sidebar icon is aria-hidden. (cherry picked from commit dabc20a) Original author: Rohan Khanderia (#30210).
) The Getting started and How this is secured cards on the Artifact Deploy page each open and close with a clickable header that had no aria-expanded, so a screen reader could not tell whether the section was open. Their chevron and title icons were also exposed to assistive tech. Both headers now carry aria-expanded, and their icons are aria-hidden. (cherry picked from commit f81f7b2) Original author: Rohan Khanderia (#30285).
The Settings model pin notice read "New chats on the <agent> agent use its own pinned model <model>, not this setting." A plain new chat does not use that pin: the New chat button starts a default-template session, and its model resolves through the template tier, which skips the crewmate pin and falls to the Default Model setting. The pin only decides the model of a chat pointed at that agent with Switch agent. Reword the notice to say that a chat pointed at the agent with Switch agent runs its pin, and that the New chat button uses this setting, so the setting no longer reads as a no-op. Update every locale. Wording only: no backend, resolver, or model-resolution change, and no new option, so the model a chat runs on is unchanged. The sibling template notice was already correct: the template's own model is what a plain new chat runs, so it is left unchanged. Update the ChatPanel notice test to the corrected copy; it fails on the old string and passes with the reword. Closes #29725 Co-authored-by: kirocrew-agent <agent@kirocrew.local> (cherry picked from commit 7720e97) Original author: Raymond Chen (#29944).
…SCII letters (#30535) PreferenceStore.add_group promised a case-insensitive match on the trimmed name, but it compared with SQLite's lower(), which folds ASCII letters only. Adding "Électronique" and then "électronique" created two groups with the same name, splitting the user's preferences. The lookup now reads the group names and compares them with str.casefold in Python, under the same lock. A regression test adds French and Russian names in two cases and checks that the second add returns the first group. Closes #30534 (cherry picked from commit aa56e70) Original author: rohnkum (#30535).
On the Skills tab, each pending skill candidate opens its review panel with a Review / Hide button that had no aria-expanded, so a screen reader could not tell whether the panel was open. The button now carries aria-expanded. (cherry picked from commit d4c35f4) Original author: Rohan Khanderia (#30489).
…0582) A dedup loser is always one document. For an artifact or upload with no file path the knowledge.dedup.collapse SEL event said loser_file=(whole-source), which reads as the whole source being removed. It now names the document by its label. Fixes #30344 (cherry picked from commit 304669f) Original author: Zezhen Xu (#30582).
A subagent waiting on its spawn approval has not launched, but its row in the subagent progress bar still drew a ticking elapsed timer next to "Waiting for your approval to start", so it read as a running agent. The row now skips the timer while the run is pending, both while it awaits approval and while it checks whether a gone approval started. Running, tool and finished rows keep it, matching the sidebar card. Closes #29812 (cherry picked from commit 7ff9157)
) With a dozen tabs open, the side panel's pinned tabs (Changes, Artifacts, Files) were squeezed to a sliver: both tab groups shrank with the same flex factor, and flex takes overflow from each item in proportion to its size, so the much wider open-tabs group dragged the pinned group down with it. The open-tabs group now gives up its width first, down to nothing when the pinned chips alone fill the strip, and is clipped to its wrapper instead of spilling a chip fragment under the + button. Its tabs stay in the "List all tabs" menu. The pinned group keeps the default factor, so at the narrowest widths it still yields after that and the trailing strip controls stay on screen. Closes #22006 (cherry picked from commit fdb1454)
…30146) is_granted() promises to fail closed on an unreadable consent store, but _read_all caught only OS errors and bad JSON. A store whose bytes are not UTF-8 (a UTF-16 save of {"enabled": true} is the likely one) raised UnicodeDecodeError instead. Every caller already catches that and keeps the socket scrubbed, so nothing was forwarded; but the consent notice and the doctor line that tell the operator how to grant it went silent, and the unreadable-store warning was never logged. Such a store now reads as no consent, like bad JSON: the socket stays scrubbed and the notice is shown. (cherry picked from commit 8e50026) Original author: Rohan Khanderia (#30146).
) The project detail Export YAML button used a native disabled while the export ran, which takes the focused button out of the tab order and drops a keyboard user to the page. Make it inert with aria-disabled and a guard instead, and pin focus and the ignored second press. (cherry picked from commit ccb85f3) Original author: Rohan Khanderia (#32279).
The task detail panel lists the tasks a finished task depended on as one sentence. It joined those translated "Task N (title)" labels with a hard-coded ", ", so German read "A, B, C" instead of "A, B und C" and Chinese got an English comma instead of "A、B和C". The list now goes through fmtList, so it follows the active language. Fixes #32165 (cherry picked from commit 251efa5) Original author: Rohan Khanderia (#32170).
…32471) A project-owned settings.local.json that appears right before Crew's O_EXCL create made the seed decline outright, while the same file present a moment earlier took the guarded exclusion. The create-race branch now tries that exclusion too, so both timings get one outcome. Bypass mode, project hooks and old adapters stay refused as before. Fixes #30918 (cherry picked from commit ca77d75) Original author: Joe Guo (#32471).
…32405) Each AWS profile row's Verify button took a native disabled while its reachability check ran. A disabled button leaves the tab order, so a keyboard press dropped focus to the page until the check settled. Mark the busy row aria-disabled and ignore a second press instead, the way the Ops Mission Control refresh does. (cherry picked from commit b94163e) Original author: Rohan Khanderia (#32405).
…32496) With Compact Empty Folders on, an empty folder's row is only a drag handle, but its header shell carries role="button" for keyboard reordering, and the base rule in index.css paints every enabled role="button" with cursor: pointer. The row looked clickable while a click did nothing. The shell now inherits the row's drag cursor. The layout editor's pane bar has the same shape: its role="button" drag handle covers the whole bar and took the same pointer over the bar's grab cursor. It inherits the bar's cursor too. Closes #30320 (cherry picked from commit 94bee0d)
…2539) The smart-search banner's Generate now button took a native disabled while its embedding run was in flight. A disabled button leaves the tab order, so a keyboard press dropped focus to the page until the run settled. Mark the busy button aria-disabled and ignore a second press instead, the way the Ops Mission Control refresh does. (cherry picked from commit e492bc0) Original author: Rohan Khanderia (#32539).
UX Review (Opus 5.5) — ✅ PASS (a BLOCK with no Trigger and no Outcome does not block)UX-level review of UX-Verdict: PASS cannot evaluate: missing screenshots for this revision. The PR changes visible copy and layout, the blind read was not performed, and the screenshots from the source PRs were never attached here. Evidence gaps
Notes
[UX-REVIEWED] 8a2c2ec Description read[DESCRIPTION-READ] 57f08d26fc26c0b8f25e4cbb5cf26d37ee5124f64e191f5fb49e0a40b5c49381 That sha256 names everything this verdict read: the PR description, plus the 3 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS No blocker: this is a cherry-pick of 25 fixes already merged on Notes
What this change shipsInventory (10 items, list capped) — 9 justifiedIntent: give 0.12.x users the 25 user-visible fixes that already landed on
[FIRST-PRINCIPLES-REVIEWED] 8a2c2ec Description read[DESCRIPTION-READ] 8cbdcfe17476d7f4d71856b6b6f19f81880b480a392507b5659c13a738927361 That sha256 names the PR description this verdict read. Recompute it with |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: these are 25 clean cherry-picks of fixes already merged on Notes
[DESIGN-REVIEWED] 8a2c2ec Description read[DESCRIPTION-READ] e0d60b48f4733a7f01b5c2e180d52b69438e63bc43ee36459b2e7f1bd476ea3f That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
Problem / Motivation
Goal: Bring 25 small fixes that merged on
mainafterv0.12.0ontorelease/0.12.0, one commit each, in a PR of their own.Why it matters
Each row is a user-visible fix from the 0.12.1 wide-bar list (Ray, 11-07). Every pick applied cleanly onto release.
Not a goal
main; nothing is changed only here.CHANGELOG.mdchange.What changed (motivation → approach → change)
Each main squash commit was cherry-picked with
-xontorelease/0.12.0@aa7dfb4aae, in merge order onmain, one commit each. Every pick applied cleanly, with no prerequisite and no hand-resolved hunk; every commit has added and removed lines identical to its main commit.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 commits carried a non-public author email or an internal co-author trailer, which the push guard refuses, so they are authored with the public identity and name their original author in the message; internal trailer lines are removed. The other 4 keep their original author.
builder-security checked: each change is the one already merged on
main; no bind, listener, new route, auth mechanism or storage grant is added.Backwards compatibility
Compatible: each change is the one already merged on
main, and its compatibility note applies unchanged.Removes nothing: no setting, key or caller is removed.
Tests
Proof: not provable -- other: a backport of main commits whose proofs ran on their own PRs
release/*, so the direct test modules were run here: every touched test file plus every test file that imports a touched module, 13 files,-n 4 --dist loadgroup. 427 passed, 14 skipped, 0 failed.tsc -p tsconfig.app.json --noEmitpasses.vitest runon every touched test file and every touched component name: 168 files, 2146 passed, 0 failed.Manual verification
N/A - each fix was verified on its own main PR. The 0.12.1 RC soak is the end-to-end check.
Screenshots / video
Why no screenshot: each frontend pick carries the same rendered change as its main PR, whose screenshots show it.
Related Issues
Backport to the 0.12 line for 0.12.1 of #30210, #30285, #29944, #30535, #30489, #30610, #30582, #30699, #31038, #31025, #30830, #31184, #31061, #31380, #31699, #31824, #31825, #30146, #32279, #32170, #32413, #32471, #32405, #32496, #32539.
Sibling sweep
Siblings: none -- each source PR carried its own sibling sweep; this PR only carries those fixes.
Pattern harvest
Not generalizable: this PR only carries fixes already on
main. The source PRs have their own harvest.Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)Contribution License Agreement
Contributions are accepted under the project's Apache-2.0 license; see the Licensing section of CONTRIBUTING.md.