Files
orca/.github/workflows/release-mac-build.yml
T
Jinjing 433df4be3c fix(runtime): prevent file watcher SIGSEGV from crashing orca serve (#8370)
* Fix crash-isolated file watcher process pool for orca-serve SIGSEGV afte

Replace the worker-thread runtime file watcher with a forked, crash-isolated
@parcel/watcher child process pool so a native FSEvents fault can no longer
take down the main/serve process, and add bounded event batching, delivery
backpressure, and quarantine-based recovery for faulty watch roots.

* Fix crash-isolated file watcher teardown and shutdown leaks

- Fault harness could throw before mkdtemp/realpath completed, skipping
  cleanup; now tracks each temp path independently and races an async
  watcher-callback error so it can't escape the try/finally unhandled.
- In-process fallback swallowed unsubscribe failures via a bare rejection
  handler that could still throw; use .catch() instead.
- Watcher process entry's cancel-subscribe handler now reuses the async
  unsubscribe path when a crawl already finished, releasing the native
  handle instead of leaking it (blocks worktree unlock on Windows).
- Runtime watcher process pool exposed no real dispose(); shutdown now
  kills pooled children so they don't outlive the main process.

* Fix disposeSlot double-iteration bug in file watcher pool teardown

Remove the unnecessary array snapshot in dispose(): disposeSlot mutates
allSlots by deleting the slot being visited, and deleting the
in-progress element during Set iteration is well-defined, so the spread
copy was dead weight left over from prior debugging.

* Fix pending file watcher installs not aborting on unsubscribe

- Local/WSL watcher installs and SSH fs.watch setup now honor the
  in-flight AbortSignal, so the last unwatch cancels a slow native
  subscribe or remote setup instead of waiting for it to finish.
- Thread signal through IFilesystemProvider.watch and SSH-backed
  file explorer watches for the same early-cancel behavior.

* Fix crash-resubscribe hangs and SSH watch teardown races in file watcher

- Add a bounded deadline for post-crash resubscription crawls so one
  stuck root quarantines instead of pinning its whole shard forever.
- Report FSEvents overflow as recoverable so delivery continues after
  a dropped-events error instead of surfacing as terminal.
- Make WSL watcher abort errors real DOMException instances so
  AbortSignal-based cancellation checks recognize them.
- Rework SSH watch registration so ownership of the shared setup
  request (not just the first caller) decides teardown, preventing
  one caller's abort from cancelling another's shared watch and
  guaranteeing exactly one fs.unwatch per registration.
- Reformat reliability-gates.jsonc arrays and refresh WSL/SSH coverage
  entries and evidence runs to match the above.

* Add CI gate to run the file-watcher SIGSEGV fault harness under Electron

- The reliability gate and release workflows (mac, Linux) previously only
  exercised the crash-isolation harness under vanilla Node, which doesn't
  catch runtime differences in the actual Electron binary that ships to
  users.
- Adds an `ELECTRON_RUN_AS_NODE=1 pnpm exec electron ...` run of the same
  harness alongside the existing Node run, so #8212's SIGSEGV-survival
  contract is proven against both runtimes before packaging.

* Add CI gate blocking Linux/macOS release packaging on watcher fault reco

Adds a contract test asserting release-cut.yml and release-mac-build.yml
run the runtime-file-watcher-fault-harness after building and before
publishing artifacts, so a regression in watcher process fault recovery
fails release packaging instead of shipping silently.

* Fix use-after-clear crash in failAllWatcherSubscriptions

Snapshot the records map before iterating, since onTerminalError
hooks can dispose the supervisor and clear `records` mid-loop,
causing a crash. Also update the matching test to assert against
the shared buildParcelWatcherIgnoreOptions helper instead of a
loose arrayContaining match.

* Fix use-after-clear crash in failAllWatcherSubscriptions

Snapshot watcher records with Array.from instead of spread, since
spread syntax over an iterator that's mutated mid-loop by
onTerminalError hooks can produce inconsistent results.
2026-07-12 18:15:15 -07:00

174 lines
7.0 KiB
YAML

name: Release macOS Build
run-name: Mac release build ${{ inputs.tag }} (${{ inputs.release_run_id }})
on:
workflow_dispatch:
inputs:
tag:
description: Release tag whose draft should receive macOS artifacts
required: true
type: string
release_run_id:
description: release-cut workflow run that requested this build
required: true
type: string
permissions:
contents: write
concurrency:
group: release-mac-build-${{ inputs.tag }}
cancel-in-progress: false
jobs:
build-mac:
if: github.repository == 'stablyai/orca'
# Why: this workflow is outside the SignPath signing run, so Blacksmith
# cannot enter Windows artifact provenance while mac notarization gets the
# faster runner.
runs-on: blacksmith-6vcpu-macos-15
timeout-minutes: 60
env:
NODE_OPTIONS: --max-old-space-size=4096
steps:
- name: Checkout
uses: actions/checkout@v6
with:
ref: refs/tags/${{ inputs.tag }}
- name: Setup pnpm
uses: pnpm/action-setup@v6
with:
run_install: false
- name: Setup Node.js
uses: actions/setup-node@v6
with:
node-version-file: package.json
cache: pnpm
# Cache the Electron binary + electron-builder tool downloads (notarytool,
# winCodeSign, nsis, squirrel, AppImage). Saves ~30-90s per job, incl. mac.
- name: Cache electron-builder downloads
uses: actions/cache@v5
with:
path: |
~/Library/Caches/electron
~/Library/Caches/electron-builder
key: electron-builder-mac-${{ hashFiles('pnpm-lock.yaml') }}
restore-keys: |
electron-builder-mac-
# Why: pnpm install triggers electron's postinstall, which downloads the
# Electron binary from GitHub release assets. GitHub's download CDN
# occasionally returns 504s that fail the whole release. Retry on
# failure so transient network errors don't require a manual re-run.
- name: Install dependencies
uses: nick-fields/retry@v4
with:
timeout_minutes: 10
max_attempts: 3
retry_wait_seconds: 30
command: pnpm install --frozen-lockfile
- name: Verify macOS signing environment
run: node config/scripts/verify-macos-release-env.mjs
env:
CSC_LINK: ${{ secrets.MAC_CERTS }}
CSC_KEY_PASSWORD: ${{ secrets.MAC_CERTS_PASSWORD }}
APPLE_ID: ${{ secrets.APPLE_ID }}
APPLE_APP_SPECIFIC_PASSWORD: ${{ secrets.APPLE_APP_SPECIFIC_PASSWORD }}
APPLE_TEAM_ID: ${{ secrets.APPLE_TEAM_ID }}
# Why: `plutil -lint` accepts duplicate plist keys, but `codesign`
# rejects duplicate entitlements after the expensive app build.
- name: Verify macOS entitlements
run: pnpm verify:macos-entitlements
# Why: telemetry's transport gate (`src/main/telemetry/client.ts:IS_OFFICIAL_BUILD`)
# requires the build identity to be the literal string `stable` or `rc`,
# substituted by electron-vite's `define` block at build time.
- name: Classify release tag for telemetry build identity
id: tag-classify
shell: bash
env:
TAG: ${{ inputs.tag }}
run: |
set -euo pipefail
# Why the optional trailing identifier: suffixed side-branch RCs
# (vX.Y.Z-rc.N.perf) are rc-channel prerelease builds — same telemetry
# identity as plain RCs.
if [[ "$TAG" =~ ^v[0-9]+\.[0-9]+\.[0-9]+-rc\.[0-9]+(\.[0-9A-Za-z]+)?$ ]]; then
identity=rc
elif [[ "$TAG" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
identity=stable
else
echo "::error::Tag $TAG does not match stable or rc pattern; refusing to build official artifact"
exit 1
fi
echo "identity=$identity" >>"$GITHUB_OUTPUT"
echo "Classified $TAG as $identity"
- name: Build app
run: pnpm build:release
env:
# Why: Vite's web build crossed Node's default old-space ceiling on
# the macOS release runner, leaving v1.4.2-rc.8 as an incomplete draft.
NODE_OPTIONS: --max-old-space-size=4096
ORCA_BUILD_IDENTITY: ${{ steps.tag-classify.outputs.identity }}
ORCA_DIAGNOSTICS_TOKEN_URL: https://www.onorca.dev/diagnostics/token
ORCA_POSTHOG_WRITE_KEY: ${{ secrets.ORCA_POSTHOG_WRITE_KEY }}
- name: Gate runtime file-watcher process isolation
run: |
# Why: #8212 is a native-process crash contract. Prove both the Node
# host and the exact Electron runtime survive SIGSEGV before packaging.
node config/scripts/runtime-file-watcher-fault-harness.mjs
ELECTRON_RUN_AS_NODE=1 pnpm exec electron config/scripts/runtime-file-watcher-fault-harness.mjs
- name: Publish release artifacts (macOS)
uses: nick-fields/retry@v4
with:
timeout_minutes: 45
max_attempts: 3
retry_wait_seconds: 30
command: node config/scripts/ensure-native-runtime.mjs --runtime=electron && ORCA_MAC_RELEASE=1 pnpm exec electron-builder --config config/electron-builder.config.cjs --mac --publish always
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
CSC_LINK: ${{ secrets.MAC_CERTS }}
CSC_KEY_PASSWORD: ${{ secrets.MAC_CERTS_PASSWORD }}
APPLE_ID: ${{ secrets.APPLE_ID }}
APPLE_APP_SPECIFIC_PASSWORD: ${{ secrets.APPLE_APP_SPECIFIC_PASSWORD }}
APPLE_TEAM_ID: ${{ secrets.APPLE_TEAM_ID }}
- name: Verify release remains draft after artifact upload
# Why: the macOS build must never be the actor that exposes a partial
# release. If an uploader or GitHub transition flips draft early, fail
# this job so release-cut never publishes the release.
shell: bash
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
TAG: ${{ inputs.tag }}
run: |
set -euo pipefail
releases_json="$(gh api "repos/$GITHUB_REPOSITORY/releases?per_page=100")"
# Why: release upload must validate the draft before it is publicly visible.
draft="$(jq -e -r --arg tag "$TAG" '
map(select(.tag_name == $tag))
| if length == 1 and (.[0].draft | type) == "boolean" then (.[0].draft | tostring) else empty end
' <<<"$releases_json")" || {
echo "::error::Release $TAG was not found in the draft-aware releases list, or its draft state was missing."
exit 1
}
if [[ "$draft" != "true" ]]; then
echo "::error::Release $TAG was published during the mac artifact upload."
exit 1
fi
# Why post-publish for macOS: electron-builder packs and uploads in a
# single `--publish always` invocation, so there is no cheap insertion
# point between pack and upload without splitting those steps.
- name: Verify telemetry constants present in app.asar
run: node config/scripts/verify-telemetry-constants.mjs