The release run re-invoked the Full native tests workflow and waited for it
before aggregating packages. That workflow already runs for every pull
request, merge group and main push, so the repeat added no coverage while
its macOS Intel job queued behind the Intel package build and became the
critical path (60 minutes end to end, packages done after 28). Aggregation
now depends on the packaging jobs only; each package still runs its own
runtime conformance and the aggregate still validates every report.
Align the 1.9.0 notes and CHANGELOG entry with the 1.8.2 format (prefixed
bilingual entries), record the session-breaker fix and the sponsor page,
and date the entry to the release day. Remove the ARM64 diagnostics
workflow and its two probe scripts from the release tree; the CI workaround
script stays because the release workflow runs it.
snapshot_session() pinned boot_attempts at the launch value whenever any
pane still had an unconfirmed AI restore target, so the 1 Hz autosave never
reset the crash-loop counter. A Codex conversation waiting at its startup
prompts (hook review, directory choice, "open in another app") therefore
counted every normal exit as a failed start, and the third launch moved the
whole workspace to session.crashed.json.
The breaker only answers "did this launch reach its first autosave"; the
restore target is still persisted through session_agent() and resumed on the
next launch. Drop the coupling and the recovery_boot_attempts field, and add
a GPUI regression that restores an unconfirmed Codex target on the third
attempt and asserts the snapshot reports zero attempts.
Include cloud settings, panel sizing, command groups, Agent integration settings and runtime input fixes from main. Bind native Codex recovery to the live pane process and preserve verified IDs; add CMD recovery choices and an explicit exit path when identity discovery fails.
The resize fixture submitted the new pane's command before its shell produced initial output. Reuse the bounded boot precondition before the first sibling probe while preserving immediate post-resize measurements and strict geometry assertions. Cover delayed and silent startup with cleanup.
Synchronize product crate versions, local dependency constraints, lockfiles, Windows resources and installer defaults. Add bilingual changes and verified contributor credits before CI packaging.
Native Codex hooks may identify a running session/group instead of the
rollout thread accepted by resume. Retain transcript paths, normalize the
thread identity, and share the parser with session discovery and probes.
Keep recovery paths when a matching provider acknowledgement omits them,
retain native identity metadata while reducing large SSH/WSL events, and
show the existing failure message when a saved target has no resume command.
Add regressions for Windows/guest paths and multiple CLI startup flows.
Apply the same provider argument rules to native and BSD process observations.
Resolve the platform temporary-directory alias in ordinary remote fixtures and
retain explicit symlink redirection tests.
Move the pending activity, input and SSH authentication decisions into their
module-owned notes under the current governance contract. Import the shutdown
draft type directly after the incoming drag-preview cleanup removes its parent
import.
Keep hook authority separate from screen observations and shell completion.
Install owned native and remote bridges, preserve durable session recovery,
and support Shift+Enter for byte-stream clients without changing native keys.
Addresses #199.
Synchronize package versions and bilingual notes with verified issue references.
Allow releases without PR contributors to omit that section, with positive and
negative format coverage as requested by the maintainer.
Keep native update process operations and default-shell resolution inside the
platform boundary. Preserve transaction authority in the updater and tighten
the platform condition budget to the measured 484.
Persist native AI recovery targets and per-pane launch contexts without treating
command submission as provider confirmation. Keep window partitions and the
active window in the durable session snapshot for immediate and scheduled updates.
Coordinate draft approval, snapshot writes, helper readiness, process exit and
installation with a committed transaction. Reverify packages and retain recovery
state on failure. Isolate acceptance installers from production integrations and
reject same-version setup that leaves the executable unchanged.
Validation: Windows GPUI suite 1552 passed; settings 55 passed; i18n 20 passed;
9 native helper scenarios; 77 installer migration checks and 41 static invariants.
Full interactive packaged upgrade acceptance remains separate.
Record the remaining PR history after its feature was integrated.
Preserve the current workflow, which runs the complete workspace and
GPUI suite through scripts/ci_native_tests.py, and all later fixes.
Retain the shared-runner cold shell startup allowance of 60 seconds;
the generated output assertions are unchanged.
Combine custom themes, updater proxies and native resource fixes with the published release source. Preserve independent interface and CJK fonts, selection visibility, saved-session startup and all existing release checks. Remove the retired terminal-view size exception after the capability extraction brought it below the standard limit.
Separate the independent macOS release metadata checks from native test jobs while retaining both architectures and the full release gate. Preserve Git dependency checkouts in the shared Cargo cache and compile tests before checking the product graph.
Persist the pending ConPTY lifetime, process snapshot and native font ownership fixes. Keep lifecycle regression tests and the Windows memory stress tool together with the ownership contracts.
Right-clicking a folder has always opened the configured default shell —
usually PowerShell — with no way to ask for a WSL distribution instead.
- `--shell <id>` takes the same ids the `shell` setting uses (`pwsh`, `cmd`,
`wsl:Ubuntu`, or an imported profile's settings id) and applies to the
first tab this process creates. It carries an id rather than a path so
every launch re-resolves the executable.
- The id travels both startup paths: the cold start (`run_shell` →
`WorkspaceStartup::NewTerminal`), and the resident hand-off, which is the
common case on Windows. `tab.new`/`window.create` gained an optional
`shell` field, so clicking the menu while Pebrel is already running adds a
tab to the existing window instead of starting a second process.
- An explicit request never silently degrades. An unresolvable id fails
(`invalid_shell`, or a startup error before any window opens), and a
`wsl:<distro>` id that the registry no longer lists still launches
`wsl.exe -d <distro>`, leaving the complaint to the distribution itself.
The `shell` setting keeps its documented fallback to the platform default.
- The winit shell rejects a shell request instead of ignoring it, so the
caller falls back to a cold GPUI start and still gets the requested shell.
The resolution lives in a new `workspace::shell_launch` module: `main.rs` has
to validate `--shell` before it opens a window, and none of it needs
`NebulaWorkspace` state — `workspace.rs` is also at its line budget, which
this keeps from growing (`architecture/file-budgets.txt`).
The installer registers one context-menu entry per registered WSL
distribution (`PebrelWsl<n>`, skipping `docker-desktop*`) and the uninstaller
reclaims them by key prefix plus the command line it wrote. `--shell` is
passed before `--working-directory`: a drive-root `%V` ends in `\"`, which
would swallow the rest of the line if the directory came first. Both the
string and the order are pinned by installer assertions.
Docs (runtime API schema + control API) and shell completions are updated.
kuddev