clippy cannot run advisory: the workflow contract forbids continue-on-error
and several crates deny lints, so it joins after the backlog is cleared.
The trigger contract now requires the default pull_request activity types
plus ready_for_review instead of forbidding a types list.
Pull requests: `lint (fmt, clippy)` and `pr-size` join the required checks
together with all five `Tests (<os>)` jobs and both `Release workspace` jobs,
so a PR can no longer merge with a red platform. Draft PRs run the three fast
platforms on every push; marking ready for review adds Intel Mac and Windows
ARM64. `pr-size` labels size/S-XL and fails above 1500 changed source lines
unless a maintainer adds `size-exempt`.
Issues: the bug form now asks for OS version, shell, program in the pane,
connection type, non-default settings and mandatory logs/screenshots, with
log paths per platform. `issue-labels` turns the dropdowns into os:/area:
labels, flags regressions, and parks reports without usable steps, evidence
or a version as needs-info with a checklist comment; an edit that fills the
gaps clears the label. config.yml pointed at the old nebula repository.
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.
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.
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.
Enable the native font backend, fit initial windows inside the usable display, and require terminal text in installed macOS screenshots. Isolate POSIX launch groups and reject runtime endpoints outside the current launch so AppImage restarts cannot reconnect to orphaned instances.
Run Weston with its X11 backend inside one private Xvfb session, retaining compositor cleanup and a cleared DISPLAY for the actual Wayland application. This supplies the wl_seat absent from the headless backend without changing conformance assertions.
Use a marked metadata-only directory for both Cargo target and intermediate outputs, reject native targets and output/config overrides, and keep Cargo.lock fixed. Preserve peer scripts while preventing stub files from entering ordinary build caches. Add ten behavior regressions to Preview preparation; these tests record commands and do not substitute for native CI.
The Intel native job completed tests and the release build but hdiutil reported ENOSPC. Drop only the completed Nebula dev artifacts, retain dependency caches, and record source size and filesystem capacity around DMG creation. Keep native packaging errors and verification authoritative.
Build completion markers from fragments so command echo cannot satisfy output waits. Assign each Windows launch gate to an owned Job Object before starting the app, then reap retained member handles during restart and close. Preserve existing history and close limits and test unrelated-process survival.
Adopt the shared source budgets, local dependency rules and tested Python
checkers. Add the architecture workflow and metadata-only cross-target
tools, and make existing filesystem test waits deterministic.
Document contributor checks in both unreleased notes without claiming
server-side branch protection or native platform validation.
kuddev