Repository navigation
fix: 0.12.1 wave 2b (6 fixes merged 11-08 19:05-19:23Z) - #33396
Conversation
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. The diff changes visible UI: the Schedule row's Run/Cancel busy state, a thread's live reply count, and reflowed copy in many locales. The PR attaches no image, so no blind read ran. Evidence gaps
Notes
[UX-REVIEWED] eb4c47e Description read[DESCRIPTION-READ] 2478b031ea62c48cd1eda6b125af5115fd2f1475df9c5d870638938e139cfd96 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 |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: these are six clean cherry-picks of fixes already merged on main. No code outside the patch still reads the retired i18n keys. Notes
[DESIGN-REVIEWED] eb4c47e Description read[DESCRIPTION-READ] 535f2b7dac1cb4fc24757d3d6978b25122033e58751d88af9b647a7ce5bd1ca7 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 |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Six fix backports from Notes
What this change shipsInventory (8 items) — 8 justifiedIntent: ship six fixes already merged on
[FIRST-PRINCIPLES-REVIEWED] eb4c47e Description read[DESCRIPTION-READ] bde53f193421b3603903e636cd972811978d111743a97a88afe2bdbbf28a5154 That sha256 names the PR description this verdict read. Recompute it with |
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: |
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: |
The Dev Fleet how-to paragraph and the Issue Radar issue and pull-request title-change lines were glued from catalog fragments around styled spans, so ja/ko/hi/bn read them in English word order. Each sentence is now one catalog message carrying the styled parts as <Trans> markup. English output is unchanged. Fixes #33210 Original author: Zezhen Xu (cherry picked from commit 8d5c189)
#33314) When an agent or another client replied in a thread, the server raised the parent message's reply count but broadcast only the reply. The page appended the reply and left the parent untouched, so a message that just got its first reply showed no "1 reply" button, and one that had replies kept its old count, until the channel was read again. The socket handler now adds one to the parent's count when a reply lands. A message the page already holds (a channel read that raced the event) is skipped, so it is neither shown nor counted twice. Original author: Rohan Khanderia (cherry picked from commit c4f4ce3)
#33316) A job row's Run and Cancel buttons took a native disabled and the label "..." while their request was out. The keyboard press that started it blurred focus to the page, so a refused run or cancel left the reader nowhere, and a screen reader announced the busy button as dots, in every language. They now stay focusable with aria-disabled and aria-busy, keep their label as the accessible name, show a spinner over it while busy (the button keeps its width, so the row does not shift), and ignore a second press. A paused job's Run keeps its native disabled. Original author: Rohan Khanderia (cherry picked from commit 88ba8b7)
…3259) On the Knowledge page, both pagers (a source group's and the flat search results') disable Next on the last page and Prev on the first, so paging onto that edge, by button or by the arrow-key shortcut, dropped keyboard focus to the page body. A shared usePagerFocus hook now moves focus to the other pager button once the new page renders, waiting when the pager is re-drawn while that page loads. Original author: Rohan Khanderia (cherry picked from commit e48fe0e)
…33340) The Connections page named the MCP entries that share a provider's stored sign-in with a hard-coded English ", " join, both in the card's "Before you disconnect" tip and in the note after a Disconnect that kept the grant. Japanese and Chinese read an English comma between the names, and English never got its "and". Format both lists with the shared fmtList helper, so each locale joins the names its own way. Original author: Rohan Khanderia (cherry picked from commit a398062)
ac76f8c to
eb4c47e
Compare
Problem / Motivation
Goal: Bring 6 fixes merged on
mainon 2026-11-08 (between 19:05 and 19:23 UTC) ontorelease/0.12.0for 0.12.1, one commit each, in a PR of their own.Why it matters
Each fix changes behaviour a 0.12.0 user can hit today. A channel thread's parent shows its reply count as replies arrive. The Schedule page's Run and Cancel buttons and the Knowledge pager keep keyboard focus while they work. The Dev Fleet how-to paragraph, the Issue Radar title-change lines, the project scaffolder's nested folder count and the Connections page's shared-grant lists now read correctly in every locale.
Not a goal
main; nothing is changed only here.CHANGELOG.mdchange.What changed (motivation → approach → change)
Each main squash commit was cherry-picked onto
release/0.12.0@109d953b86, one commit each inmainmerge order, with the(cherry picked from commit …)trailer. The repository's leak guard requires the public identity, so every commit is authored as Ray Xu and names its original author on anOriginal author:line.8d5c1895e0194d71de11e125862432a3840ae863c4f4ce394377340ae1fe88ba8b7db334c0c1fca9e48fe0ec03ba0bc4958ea398062bc0eb4c47ebc7Hand-resolved hunks:
website/src/i18n/locales/en.jsonanden-XA.json(Dev Fleet block): keeps 0.12'srowskey and adds onlyrow_actions_how_to.main's adjacentrows_count_*/rows_filtered_count_*keys come from another change and are not brought in.website/src/pages/connections/ConnectionsPage.tsx:15-16: import lines resolved tofmtListfrom../../i18n/formatandfindReportfrom../../utils/errorReport(0.12 does not importparseErrorCodethere).Files left out of the picks: the screenshot capture helpers
website/capture/channel-live-reply-count.html,website/capture/channel-live-reply-count.tsxandwebsite/scripts/capture-channel-live-reply-count.mjs(from #33314) andwebsite/scripts/capture-33210-styled-name-sentences.mjs(from #33327). They only produced the main PRs' screenshots and ship nothing.One file under
.github/changes: #33316 removessrc/hooks/useCronActions.tsfrom.github/coverage-baselines/frontend.txt, as onmain.Left out of this batch:
queue_drain.py:1103in core names Spec Builder'sspec_decision_delivery_id(s)meta keys, againstno-new-builtin-apps. It waits for 0.12.2.kas_agents.py/kas_permissions.pyhunks callcanonical_tools_settings,as_authored,projection_acceptsandtrusted_agents_withheld, which exist only onmain. That is a wide KAS prerequisite chain, so it waits.CrewmateStarButton.tsx, which the feature #30958 added onmainand 0.12 does not have.env_node_pinsmodule and eight source conflicts inenv.py,service/common.pyandservice/linux_install.py. Too wide for a backport batch.Backwards compatibility
None: each fix keeps its main PR's compatibility notes; nothing here changes stored data or config beyond what those PRs already shipped on
main.Tests
Proof: not provable -- other: a backport of main commits whose proofs ran on their own PRs
eb4c47ebc7): vitest on the 8 changed test files, 265 passed.vitest run src/i18n/: 71 files, 1,197 passed.check-i18n-strings.mjsandi18n-check.mjs(I18N_BASE_REF=109d953b86) pass.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: the frontend picks carry the same rendered change as their main PRs, which show it.
Related Issues
Backport to the 0.12 line for 0.12.1 of #33327, #33257, #33314, #33316, #33259, #33340.
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.