StopCancelled carries no backgroundTasks, so a cancelled Grok turn settled
done while the shells and subagents it left running kept going. Track
Grok's own task ids per pane and settle the cancelled row on each task's
own end.
* fix(wsl): session history reads never boot a stopped WSL distro (E9)
Transcript and session-history reads touched \\wsl.localhost shares, and ran
`wsl.exe --exec` probes, for any distro the root list named. Those touches are
the OpenCode reader, its setup, HOME and realpath probes, and the session
scanner and search index. Nothing re-checked that the distro still ran. A
`wsl --terminate` mid-scan, or a list taken while the distro was still
stopping, let background work boot it again.
Every UNC transcript touch (stat, readdir, open, access, readFile) and every
scanner `--exec` now asks a settled running read first. That read is no older
than 5 s, because the scanner process hears of no stop by itself. A user act on
the distro still passes.
* fix(wsl): opening a chat starts its stopped distro; scanner hears main's stop notes
A user-opened chat, a history page and an AI Vault resume start the
transcript's stopped WSL distro once, as a user act, so its history
shows; watch and scan reads stay background. Main forwards WSL stop and
wake notes to the session-scanner process so a terminate main saw is
honoured there at once.
* fix(wsl): type the scanner note channel structurally; test the resume wake
Keeps node:child_process out of the new scanner-notes module (the
process-host import check) and adds a test that fails if an AI Vault
resume stops starting the session's WSL distro.
L1: the orcad liveness supervisor's heartbeat/verify reads, its recovery gate, the no-boot
reconnect check and the connector's passive probe now take a settled read of their own distro, so
a list probed while the distro is still stopping cannot admit a holder spawn, identity read or
reconnect that would boot it again.
L2: a Git cancel/timeout's guest kill, the WSL move's process table and each step of the Codex
session bridge chain check the distro first. A confirmed stop counts as the group gone / an empty
table / the end of the chain; an unanswered read proves nothing and spawns nothing.
A slot whose startup qualification marker matches (same artifacts, runtime, libc, kernel and
launch env, last launch reached readiness) now also skips the synchronous node-pty load probe
child (~0.1 s on the WSL lane). The marker is now recorded only when that launch's probe
passed cleanly; schema 2 so earlier markers, which did not promise that, re-qualify once.
Key MermaidViewer's scroll cache with the pane-scoped editorViewStateKey,
matching IpynbViewer, so split panes showing the same .mmd file no longer
overwrite each other's saved diagram position.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Zoom and resize re-renders reuse the same key string instead of rebuilding
and comparing it, which matters for base64-backed images (diff, conflict
review, fallbacks) where the key embeds the whole payload.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Key the colorize effect on joined content and line count instead of the
lines array identity, so re-renders with an equal fresh array (PR comment
code context) no longer re-run monaco.editor.colorize. Check cancellation
before colorizing so excerpts replaced or unmounted while a lazy tokenizer
loads skip the work.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Monaco's CopyPasteController claims paste on the editor container in the
capture phase before Orca's listener, so Ctrl/Cmd+V bypassed the 16 MiB
cap. Listen on the editor shell instead, scoped to this editor's subtree
and text focus, and only reject over-cap pastes; smaller pastes keep
Monaco's atomic native paste. Remove the DOM chunked-paste plumbing
(save deferral, isApplyingLargePasteRef) that could never run.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Prune file/diff read generations only when a tab goes from open to closed,
replacing the append-only known-open-ID set, and fence the restored-owner
migration so a late result cannot discard a newer read's generation.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Upload up to four files concurrently within the existing 2 MiB in-flight budget, sizing each file's window from its size. A per-file failure aborts siblings and surfaces the first error after the SFTP channel closes.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Two clients hold separate grants for one file, and the desktop editor saves through its own guarded path, so concurrent writers could all pass their content check before any of them replaced the file, and the last one silently won. Run the artifact check-to-rename under the same canonical-path/inode lock that guarded editor saves already use; the later writer now sees the earlier one's bytes and is refused.
* feat(mobile): add UI localization infrastructure
Mobile follows the phone's language (or an in-app choice under Settings > Appearance)
across the same 11 languages as desktop, switching live like the theme does.
- i18next with desktop's translate(key, fallback, options) call shape; catalogs in
mobile/src/i18n/locales load lazily, English fallbacks stay inline
- Hermes lacks Intl.PluralRules: load the FormatJS polyfill with the 11 languages' rules
- OTA pages get the shell's language through an optional bridge init field
- iOS permission prompts localized; iOS and Android 13+ per-app language pickers
- verify:mobile-localization and a coverage ratchet run in lint, check:changed and CI
* fix(mobile): mount pages after the newest language request; tighten localization gates
- Wait for the latest page language activation, not a superseded one
- Keep literal fallbacks when a key also has a dynamic one
- Count headerTitle as visible copy; select the mobile check when the runtime catalog generator changes
* feat(mobile): repaint themed components on language changes
Components that read the theme already subscribe for repaint; a language switch
repaints them too, so labels built by helpers during render follow it. RPC
recordings load the real i18next and JSON catalogs.
* fix(mobile): scope the translate() gate to render bodies and memos; prune stale keys on sync
Handlers and effects may call translate() when they run. sync:mobile-localization now
removes keys no code references from every catalog.
* fix(mobile): use node: imports in the locale config plugin; pass uiLocale in bridge hook test
* chore(mobile): refresh the localization coverage baseline
* fix(mobile): follow render callbacks in the translate() gate; keep used plural siblings on prune
Also let anonymous default-exported screens count as components, flag any plural
family a locale started, and mark JSON modules as ES modules in the recording loader.
* fix(mobile): treat async and named local callbacks as handlers in the translate() gate
* chore(mobile): refresh the localization coverage baseline
* chore(mobile): refresh the localization coverage baseline
---------
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Cap each splice direction at 4,096 queued frames (closes 4429 with
trigger queue-frame-limit) and charge a 64-byte per-entry overhead
against the shared process queued-byte budget so empty frames are not free.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
* fix(i18n): localize plural counts, sentence fragments and hardcoded copy in editor, chat, skills, stats, onboarding and tips
Refs #32207
* chore(i18n): resync runtime English catalog after merging main
* test(i18n): let mocks interpolate counts and keep real i18n exports
---------
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Handle 'error' on accepted CDP proxy sockets so an invalid frame (bad
UTF-8, bad opcode) closes that client instead of crashing the main
process. Validate parsed command envelopes: null params/sessionId stay
treated as absent, malformed bodies with a numeric id get an
'Invalid CDP request' error reply, and id-less garbage is dropped.
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
* fix(i18n): localize sidebar, status bar, tab bar and workspace cleanup copy
Replace fragment-concatenated sentences, separate singular/plural keys and
raw English in the S2 renderer areas (sidebar, status-bar, tab-bar,
tab-group, workspace-cleanup, new-workspace, activity, dashboard,
terminal-pane, cmd-j) with whole-sentence keys and CLDR plural families,
and add translations for all built-in locales.
Refs #32207
* fix(i18n): subscribe DashboardAgentRow to UI language changes
* test: keep real i18n exports in SmartWorkspaceNameField mocks for Trans rows
---------
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
Paired phones, web clients and the CLI poll reads nobody is looking at:
history, submodule status, ignored paths, the conflict operation, the
in-flight mutation probe, upstream status and link-existence checks.
Their desktop IPC twins already refuse a stopped distro; these did not, so
a client left open booted the distro again after the user terminated it.
Each now asks the settled running gate first and refuses with
WslDistroStoppedError, which reaches every client version as the existing
runtime_error.
User acts stay ungated: git.status (a viewing client's implicit wake),
diffs and compares (opened views), file reads and directory listing
(taps and browsing), and stat (the download name check).
* fix(mobile): retain renderer-owned browser tabs while unmounted
Adapt the community browser ownership fix from #14792 to the current runtime modules. Preserve renderer epochs and route mobile browser close through its owner.
Co-authored-by: Bing.Z <zzb@gxsmjx.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
* test(mobile): cover browser ownership lifecycle races
* test(runtime): cover browser identity collisions and stale retirement
* fix(mobile): retain browser group membership after renderer reconnect
* fix(mobile): drop renderer browser rows once the desktop window closes
In serve mode the offscreen reconcile no longer checked for a window, so a
closed window's browser rows stayed listed with no page behind them. Also
drop the legacy :headless-merge: ownership fallback, which nothing produces.
* fix(mobile): list a desktop webview's page once in serve mode
The bridge inventory also lists an attached window's webviews. Only the
offscreen reconcile filtered those out, so hydrate added a second row for
the same page, and closing it unregistered the desktop's guest. Filter in
the shared builder instead.
* fix(mobile): list only pages the serve browser backend created
With a desktop window attached in serve mode the bridge inventory also
lists the window's background webviews, which the renderer does not
publish. Those became offscreen rows, and closing one unregistered the
desktop's guest. Ask the backend which pages it owns.
* style(mobile): keep projection layout file under the line cap
* fix(mobile): keep the serve publication epoch when browser pages change
* test(mobile): give browser backend doubles ownsTab and drop removed notifier field
* fix(mobile): republish browser rows when the desktop window closes and dedupe serve rows
* fix(mobile): rebase detached publications only for retained offscreen pages
* fix(mobile): advance snapshot versions when the desktop window closes
* test(mobile): name the offscreen epoch ownership tests after what they cover
---------
Co-authored-by: Bing.Z <zzb@gxsmjx.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
Replace sentence fragments, manual non-family plurals, item(s) copy and
hard-coded English in the right sidebar, GitHub item dialog, PR page,
GitLab item dialog and diff-comment surfaces with whole-sentence keys and
CLDR plural families, and add translations for all built-in locales.
Refs #32207
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
* fix(i18n): localize remaining renderer plurals, fragments and hardcoded copy
Convert manual one/many keys and value0 counts to CLDR plural families,
replace fragment concatenations with whole-sentence keys or Trans, and route
hardcoded English labels, errors and compact time units through the UI locale.
Refs #32207
* fix(i18n): re-render memoized agent state dots on language change
---------
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
With "Remote server workspaces" set to "Server (streamed)", a paired server
that has no way to stream pages (an orcad `orca serve`: its browser engine
offers no screencast) could not take the desktop's new tabs. The desktop
fell back to a plain desktop tab the server never heard of: it browsed from
the desktop's network and never appeared in `orca tab list`, so agents could
not see or drive the page the user was on.
The setting is a rendering preference. When the server cannot stream, the
desktop now renders the page itself as a page the server owns (the existing
client-hosted path), so traffic still goes through the server and the tab is
listed and drivable from the server's CLI. Servers that can stream keep
rendering on the server; an explicit server placement is unchanged.
Refs #35156
* fix(wsl): account sync and usage previews never boot a stopped WSL distro (E9)
Four account paths touched WSL distros in the background:
- Every launch mirrored ~/.codex/config.toml into each WSL Codex account home over UNC.
- Inactive Codex usage previews read a WSL home that nobody had checked.
- The startup Claude publish and inactive Claude previews trusted an unsettled
running list.
Any of these booted a stopped distro, or one still stopping. The mirror now skips
WSL homes not seen running after any stop; launch preparation syncs them before
use. The other three use the settled running gate.
* fix(wsl): the Codex mirror trusts only a running list from the last 60 s
WSL's own idle stop is reported by nothing, so an old list must not
vouch for a distro. Drop the inert user-act clause: select and sign-in
do not run as WSL user acts.
* feat(wsl): a WSL bridge shell reports its folder, so restores follow cd
Bridge bash and zsh send no OSC 7, so the terminal daemon's emulator and every checkpoint kept
the spawn folder until a Sleep read the shell's folder in the distro. A terminate, a quit or a
timed-out read reopened the pane at the project root (verify-18 e). The bash rcfile and zsh
wrapper now report the folder at each prompt where it changed, only inside WSL, percent-encoded;
the daemon already reads OSC 7 into a WSL pane's UNC path.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(wsl): re-send the folder report every prompt, and prove the daemon bridge sources it
* fix(wsl): keep the zsh folder report's locals in Orca's namespace; update wrapper snapshots
---------
Co-authored-by: Claude <noreply@anthropic.com>
* fix(mobile): clear terminal output only after the computer accepts Clear
The phone erased its terminal before sending terminal.clearBuffer and
reported success on any reply, refusal included. Wait for the host's
{ clear: { cleared: true } } reply, fail immediately when disconnected so
the request cannot replay after reconnect, ignore late replies after the
client or terminal changed, and resubscribe a subscribed terminal so its
snapshot (including output that arrived during the request) replaces the
old screen. An unsubscribed tab draws its snapshot when it next subscribes.
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
* test(mobile): return the subscribe result from the gesture recording adapter
* fix(mobile): keep accepted clears when the terminal view re-renders
The WebView handle is rebuilt on every render and a relay/direct cutover
bumps the connection generation without changing the host, so both
checks treated accepted clears as stale: the phone neither reloaded the
output nor showed a toast. Only a replaced client is stale now.
---------
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
* chore(wsl): keep the Sleep folder and bridge-exit lines in the local trace
A packaged build keeps no console, so #35150's Sleep folder line and the WSL bridge exit verdict
were invisible on a device. Each now also goes to main.trace.ndjson as a span with the line and
its facts as filterable fields.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(wsl): keep session paths and distro names out of the persisted WSL trace
---------
Co-authored-by: Claude <noreply@anthropic.com>
* fix(mobile): drop voice settings replies older than the latest change
An older save that replies last could put the previous setting back on screen. Ignore mutation replies and errors that a newer change has superseded.
* fix(mobile): keep the Voice screen on the connected desktop
Key the screen by host so a late reply from a desktop the phone switched away from cannot land on the newly connected desktop's settings.
* fix(mobile): re-read Voice settings when an older save succeeds late
The host finishes a model delete only after removing its files, so a newer change can reply first with the model still selected. Dropping the older reply left the deleted model shown as in use; re-read instead.
* fix(mobile): keep Voice model errors and the screen through reconnects
A newer setting change does not replace a failed model select, download or delete, so show those errors again. Key the screen on the last connected desktop so a reconnect of the same desktop no longer remounts it.
* test(mobile): pin Voice model-delete errors after a newer change
* fix(mobile): re-read Voice settings after any failed save
* fix(mobile): re-read Voice settings only after every pending change settles
A read sent while a model delete is still running can return the deleted model after the delete's own reply. Skip reads while a change is pending, and re-read once when all settle if any reply was skipped.
* fix(mobile): keep a failed Voice model action's error on screen
A lone failed select or delete changed nothing on the phone, so it no longer starts a re-read that clears its error. A failed switch still re-reads to undo its optimistic flip.
* fix(mobile): let Voice reads clear only their own load errors
A re-read after a failed or superseded change cleared the change's error before the user could read it. Keep load and change errors apart; re-read after any failure again, which also undoes a failed switch flip.
---------
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
* Nest chat-UI orchestration workers under the agent that started them
Main's lineage projection only walked terminal panes, so workers running
in the chat UI never got a parent link and listed flat in the sidebar.
It now also projects chat workers, keyed by their visible chat row, and
resolves a chat creator or coordinator to that chat's visible row
instead of its handle or private mailbox pane key.
* Fix public worker lineage across assignment and row identities
* Complete terminal lineage provenance and bound active session selection
* Repair lineage test runtime boundaries and SQLite counters
* Batch public lineage verification after exact recency selection
* Use checked runtime values in lineage integration tests
* Document the trusted Dispatch SQL row boundary
* Keep lineage integration imports within their test programs
* Load native lineage fixtures before test execution
* test: complete lineage inputs in runtime and sidebar fixtures
* test: seed benchmark facts from active dispatches
* test: replace partial orchestration DB casts with typed fixtures
* test: complete stale-leaf mailbox fixture session input
* fix(test): render worker lineage fixture with production list
The chat mic only got isRecording/isProcessing, so it looked idle while starting; it now shows the live look the terminal mic does. The terminal status row says 'Release mic to stop' in hold mode, both mics expose role and busy state, and one live region announces Starting mic / Listening / Processing.
Fixes#35422
* fix(mobile): preserve file drafts across refresh and reject stale reads
Overlapping preview reads overwrite current content, and refreshed metadata or equivalent route objects revoke dirty editor state. Keep document-owned drafts editable and accept only the current load.
* fix(mobile): prevent duplicate artifact saves before rerender
* refactor(mobile): drop the dead dirty-draft guard and share preview test setup
---------
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
* perf(wsl): fold the first libc probe into the WSL host facts exec
The WSL connect's facts exec now also runs the libc probe's first command after its own marker.
resolveOrcadRemoteContext takes it (with the login home) as a known fact, answering that one
probe; the fallbacks still run over the transport if it proves nothing. One fewer serial guest
round trip per connect; SSH passes no known facts, so its probes are unchanged.
* fix(wsl): run the libc fallbacks through the default exec
With a folded first answer, the remaining libc probes used the fenced orcad runner; they now
use the same plain execCommand the default probe does, so they match main exactly.
* fix(mobile): keep dictation recording through a reconnect
The host tied a phone's dictation to its socket, so any reconnect cancelled
it, and the phone ended the recording as soon as the connection dropped.
The host now owns the session by device and waits a grace period for the
phone to come back. The phone keeps recording through the reconnect and
keeps the text transcribed so far if the connection does not return.
Fixes#35211
* fix(mobile): bound dictation reconnect waits and release retired relay waiters
* fix(mobile): resend an unsent dictation finish after a connection move and send the text so far only when it changes
* fix(mobile): hold dictation requests through a suspended relay and bound the start wait
* fix(mobile): recognise a suspended relay client across the page bridge
* refactor(dictation): refresh one abandon timer, report text by version and share the reconnect wait
* fix(mobile): resend dictation text each second and never wait offline for a cancel
Kelvin Amoaba
Neil
OrcaWinandOrcaWin
Jinjing
Brennan Benson