Files
orca/.github/workflows/hourly-mac-build.yml
T
NeilandOrca cd2b62ed14 feat(updater): name hourly releases by version, build number, time, and sha (#11817)
* feat(updater): name hourly releases by version, build number, time, and sha

Hourly releases were titled with their raw tag
(`v1.4.163-hourly.202607312054`), which reads as one opaque digit run and
does not say which commit it came from.

Title them `1.4.163 • 01 • 07-31 13:54 • e698241` instead, and show that
same string in the in-app build picker by having the picker render the
release's stored name rather than deriving its own label. Composing it in
one place means the two surfaces cannot drift.

The build number is monotonic across the channel. It is read as the
highest number already in use rather than as a count of releases: the
prune step trims to 72, so a count would roll backwards after three days
and reissue numbers. Drafts count toward it — unlike in the freshness
check, which asks whether a commit shipped, this asks whether a number is
free, and a stranded draft still holds one.

Times are Pacific while the tag's stamp stays UTC. The stamp is a sort
key and a local one would repeat an hour at every DST fall-back, making
two distinct builds compare equal; the title is only ever read.

* fix(updater): fail the hourly build when the release name is missing

The workflow checks out `ref: main`, but a workflow_dispatch runs the
workflow file from whatever branch was dispatched. A branch that edits
this step while main still carries the old script produces an empty name
and an untitled release — silent, and only visible once someone opens the
releases page. Verified by hitting exactly that on run 30665586904.

Co-authored-by: Orca <help@stably.ai>

---------

Co-authored-by: Orca <help@stably.ai>
2026-07-31 14:29:20 -07:00

368 lines
17 KiB
YAML

name: Hourly macOS Dev Build
# Why: gives developers a signed macOS build of main every hour that the in-app
# updater can install directly, without waiting for an RC cut.
#
# Deliberately narrow scope:
# - macOS only. Other platforms keep using RC/stable.
# - No tests, no lint, no e2e. This channel trades safety for latency; PR CI
# and release-cut remain the gates that matter.
# - Signed but NOT notarized. Squirrel.Mac validates the replacement bundle's
# signature, not its notarization, so in-place updates still work while the
# ~10min notary round trip is skipped 24x a day.
#
# Artifacts publish to stablyai/orca-hourly, never to stablyai/orca: the main
# repo's releases atom feed exposes only its 10 newest entries, so 24 hourly
# tags a day would evict every stable/RC entry and break updates for real users.
#
# GITHUB_TOKEN is scoped to this repo and cannot publish there, so writes use a
# GitHub App installed on orca-hourly with Contents: Read and write. Its private
# key does not expire, unlike a PAT — nothing here needs yearly rotation, and the
# credential belongs to the org rather than to whoever created it.
#
# Provision the two secrets with `bash config/scripts/setup-hourly-release-token.sh`:
# HOURLY_RELEASE_APP_ID the App's numeric id
# HOURLY_RELEASE_APP_PRIVATE_KEY the App's .pem private key
#
# Installation tokens live one hour, which is ample here: this job skips tests,
# notarization, and Windows signing entirely, so a run is pack + upload and lands
# well inside that. If one ever did overrun, the release is still an unpublished
# draft at that point, so nothing user-visible breaks.
on:
schedule:
# Top of every hour. Skipped automatically when main has not moved.
- cron: '0 * * * *'
workflow_dispatch:
inputs:
force:
description: Build even if main has not moved since the last hourly
required: false
default: false
type: boolean
permissions:
contents: read
concurrency:
group: hourly-mac-build
cancel-in-progress: false
env:
HOURLY_REPO: stablyai/orca-hourly
# Keep ~3 days of history so a regression can be bisected across a weekend.
HOURLY_RETAIN_COUNT: 72
jobs:
build-hourly-mac:
if: github.repository == 'stablyai/orca'
runs-on: blacksmith-6vcpu-macos-15
# Why 120: it must exceed the worst case the retry budgets below can produce
# (install 3x10 + publish 2x25 = 80, plus ~30 for checkout/build/verify), or
# the job is killed mid-retry and no cleanup step runs at all. A typical run
# is far shorter — there is no notarization, Windows signing, or test phase.
timeout-minutes: 120
env:
NODE_OPTIONS: --max-old-space-size=4096
steps:
- name: Checkout
uses: actions/checkout@v6
with:
ref: main
fetch-depth: 0
# Why: this job only reads stablyai/orca and never pushes; every write
# goes to the hourly repo through a minted App token passed by env.
# Not persisting the checkout credential shrinks the blast radius if a
# build step is compromised (zizmor: artipacked).
persist-credentials: false
- name: Mint hourly repo token
id: app_token
uses: actions/create-github-app-token@v2
with:
app-id: ${{ secrets.HOURLY_RELEASE_APP_ID }}
private-key: ${{ secrets.HOURLY_RELEASE_APP_PRIVATE_KEY }}
owner: stablyai
repositories: orca-hourly
# Why: main is often idle overnight. Rebuilding an unchanged commit burns a
# runner hour and adds a redundant tag to the retention window.
- name: Check whether main moved since the last hourly
id: freshness
shell: bash
env:
GH_TOKEN: ${{ steps.app_token.outputs.token }}
FORCED: ${{ github.event_name == 'workflow_dispatch' && inputs.force }}
run: |
set -euo pipefail
head_sha="$(git rev-parse HEAD)"
echo "head_sha=$head_sha" >>"$GITHUB_OUTPUT"
if [[ "$FORCED" == "true" ]]; then
echo "should_build=true" >>"$GITHUB_OUTPUT"
echo "Forced dispatch; building $head_sha."
exit 0
fi
# The previous hourly records its source commit in the release body.
# Drafts are excluded: an unpublished leftover never shipped, so treating
# it as "the last build" would skip a build that never actually happened.
last_body="$(gh release list --repo "$HOURLY_REPO" --limit 20 --json tagName,isDraft \
--jq 'map(select(.isDraft | not)) | .[0].tagName // empty' 2>/dev/null || true)"
if [[ -z "$last_body" ]]; then
echo "should_build=true" >>"$GITHUB_OUTPUT"
echo "No prior hourly release found; building $head_sha."
exit 0
fi
last_sha="$(gh release view "$last_body" --repo "$HOURLY_REPO" --json body \
--jq '.body | capture("commit `(?<sha>[0-9a-f]{7,40})`") | .sha' 2>/dev/null || true)"
if [[ -n "$last_sha" && "$head_sha" == "$last_sha"* ]]; then
echo "should_build=false" >>"$GITHUB_OUTPUT"
echo "main is unchanged since $last_body ($last_sha); skipping."
else
echo "should_build=true" >>"$GITHUB_OUTPUT"
echo "main moved to $head_sha (last hourly built $last_sha); building."
fi
- name: Setup pnpm
if: steps.freshness.outputs.should_build == 'true'
uses: pnpm/action-setup@v6
with:
run_install: false
- name: Setup Node.js
if: steps.freshness.outputs.should_build == 'true'
uses: actions/setup-node@v6
with:
node-version-file: package.json
cache: pnpm
- name: Cache electron-builder downloads
if: steps.freshness.outputs.should_build == 'true'
uses: actions/cache@v5
with:
path: |
~/Library/Caches/electron
~/Library/Caches/electron-builder
key: electron-builder-mac-${{ hashFiles('pnpm-lock.yaml') }}
restore-keys: |
electron-builder-mac-
- name: Install dependencies
if: steps.freshness.outputs.should_build == 'true'
uses: nick-fields/retry@v4
with:
timeout_minutes: 10
max_attempts: 3
retry_wait_seconds: 30
command: pnpm install --frozen-lockfile
# Why: signing is what makes an hourly installable over an existing Orca, so
# a missing cert must fail here rather than after a 20-minute build.
- name: Verify macOS signing environment
if: steps.freshness.outputs.should_build == 'true'
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 }}
- name: Compute hourly version
id: hourly
if: steps.freshness.outputs.should_build == 'true'
shell: bash
env:
GH_TOKEN: ${{ steps.app_token.outputs.token }}
run: |
set -euo pipefail
# Why read the highest number off existing titles rather than counting
# releases: the prune step trims to HOURLY_RETAIN_COUNT, so a count would
# roll backwards after three days and reissue numbers already in use. The
# maximum is always among the retained builds, because pruning drops the
# oldest. `capture(...)?` swallows the non-match on legacy titles, which
# were the raw tag, so the series simply starts at 01.
#
# Why drafts count here but not in the freshness check: that check asks
# "did this commit ship", where a draft is a no. This one asks "is the
# number free", where a stranded draft still holds one.
last_number="$(gh release list --repo "$HOURLY_REPO" --limit 200 --json name,isDraft \
--jq '[.[] | (.name // "") | capture(" • (?<n>[0-9]+) • ")? | .n | tonumber] | max // 0')"
build_number=$(( last_number + 1 ))
echo "Hourly build number $build_number (previous high: $last_number)"
ORCA_HOURLY_BUILD_NUMBER="$build_number" node config/scripts/hourly-build-version.mjs \
>"$RUNNER_TEMP/hourly-identity.txt"
# Why check rather than trust: the checkout above pins `ref: main`, but a
# workflow_dispatch runs this file from whatever branch was dispatched. A
# branch that edits this step while main still has the old script yields
# an empty name and an untitled release — silent, and only visible once
# someone opens the releases page. Fail here instead.
if ! grep -q '^name=' "$RUNNER_TEMP/hourly-identity.txt"; then
echo "::error::hourly-build-version.mjs emitted no release name; this workflow and main's copy of the script are out of sync."
exit 1
fi
cat "$RUNNER_TEMP/hourly-identity.txt" >>"$GITHUB_OUTPUT"
- name: Build app
if: steps.freshness.outputs.should_build == 'true'
run: pnpm build:release
env:
NODE_OPTIONS: --max-old-space-size=4096
# Why: hourly builds are not an official channel — telemetry's transport
# gate accepts only 'stable' or 'rc', so leaving this unset keeps them
# silent, which is correct for unvetted dev artifacts.
ORCA_DIAGNOSTICS_TOKEN_URL: https://www.onorca.dev/diagnostics/token
- name: Create hourly release
id: release
if: steps.freshness.outputs.should_build == 'true'
shell: bash
env:
GH_TOKEN: ${{ steps.app_token.outputs.token }}
TAG: v${{ steps.hourly.outputs.version }}
NAME: ${{ steps.hourly.outputs.name }}
SHA: ${{ steps.freshness.outputs.head_sha }}
run: |
set -euo pipefail
# Kept at 12 even though the title shows 7: the freshness check above
# parses this back out of the body to decide whether main has moved.
short_sha="${SHA:0:12}"
# Why create it up front: electron-builder then uploads into a known tag
# rather than inferring one from package.json.
#
# Why --draft: everything between here and the manifest check is a window
# where the release exists but has no installable assets. A draft is
# absent from the releases list and from listReleaseBuilds, so a job that
# dies in that window — including a hard kill by the job timeout, which
# runs no cleanup step at all — leaves something invisible rather than a
# tag the picker offers and the download 404s on. It is flipped live only
# after the manifest is verified.
gh release create "$TAG" \
--repo "$HOURLY_REPO" \
--title "$NAME" \
--draft \
--notes "Automated hourly macOS dev build from commit \`$short_sha\`.
Built from [\`stablyai/orca@$short_sha\`](https://github.com/stablyai/orca/commit/$SHA).
**Unvetted.** No tests ran. Signed but not notarized — installable
through Orca's in-app updater, but a manual download will be
Gatekeeper-quarantined."
echo "tag=$TAG" >>"$GITHUB_OUTPUT"
- name: Publish hourly macOS artifacts
if: steps.freshness.outputs.should_build == 'true'
uses: nick-fields/retry@v4
with:
# Why lower than the release pipeline's 3x45: that budget is sized for
# notarization and Windows signing, neither of which runs here. An attempt
# is pack + upload only.
timeout_minutes: 25
max_attempts: 2
retry_wait_seconds: 30
command: node config/scripts/ensure-native-runtime.mjs --runtime=electron && ORCA_MAC_HOURLY=1 pnpm exec electron-builder --config config/electron-builder.config.cjs --mac --publish always
env:
# Why: electron-builder's github publisher targets the repo named in the
# config; the token must therefore carry write access to orca-hourly.
GH_TOKEN: ${{ steps.app_token.outputs.token }}
ORCA_HOURLY_BUILD_VERSION: ${{ steps.hourly.outputs.version }}
ORCA_BUILD_COMMIT: ${{ steps.hourly.outputs.commit }}
CSC_LINK: ${{ secrets.MAC_CERTS }}
CSC_KEY_PASSWORD: ${{ secrets.MAC_CERTS_PASSWORD }}
APPLE_TEAM_ID: ${{ secrets.APPLE_TEAM_ID }}
# Why: the updater resolves a tag, then fetches latest-mac.yml from it. A
# release missing that manifest is a tag the picker offers and the download
# 404s on, so fail loudly instead of leaving a broken entry.
- name: Verify update manifest published
if: steps.freshness.outputs.should_build == 'true'
shell: bash
env:
GH_TOKEN: ${{ steps.app_token.outputs.token }}
TAG: ${{ steps.release.outputs.tag }}
run: |
set -euo pipefail
assets="$(gh release view "$TAG" --repo "$HOURLY_REPO" --json assets --jq '.assets[].name')"
echo "Published assets:"
echo "$assets"
# Why exit 1 without deleting here: the release is still a draft, so it is
# already invisible to users, and the failure handler below owns cleanup.
# Deleting inline under `set -e` would also let the delete's exit code
# preempt this explicit failure.
for required in latest-mac.yml; do
if ! grep -qx "$required" <<<"$assets"; then
echo "::error::Hourly draft $TAG is missing $required; the updater could not install it."
exit 1
fi
done
if ! grep -q '\.zip$' <<<"$assets"; then
echo "::error::Hourly draft $TAG has no ZIP artifact for the updater to download."
exit 1
fi
# Why this is the last mutating step: publishing the draft is what makes the
# build visible to listReleaseBuilds. Doing it only after the manifest check
# means the picker can never offer a release whose assets are incomplete.
- name: Publish the verified release
id: publish_live
if: steps.freshness.outputs.should_build == 'true'
shell: bash
env:
GH_TOKEN: ${{ steps.app_token.outputs.token }}
TAG: ${{ steps.release.outputs.tag }}
NAME: ${{ steps.hourly.outputs.name }}
run: |
set -euo pipefail
# --title again: electron-builder resolves this draft by tag and may
# rewrite its title on upload. Re-asserting here means the name the
# picker reads is the one composed above, whatever it did in between.
gh release edit "$TAG" --repo "$HOURLY_REPO" --draft=false --prerelease --title "$NAME"
echo "Published $TAG as \"$NAME\""
# Why: a draft left behind by a failed publish is invisible to users but still
# holds its tag name, so the next run for the same minute would collide.
#
# Why it is gated on publish_live not having succeeded: a later failure (the
# prune step) must not delete a release that already went live and that users
# may already be installing. A job killed by the outer timeout runs no steps
# at all — which is exactly why the release stays a draft until verified.
# Why cancelled() too: a run stopped from the Actions UI is not a failure(),
# so without it a manual cancel mid-publish would strand the draft.
- name: Discard the draft release on failure
if: >-
(failure() || cancelled()) && steps.release.outputs.tag != '' &&
steps.publish_live.outcome != 'success'
shell: bash
env:
GH_TOKEN: ${{ steps.app_token.outputs.token }}
TAG: ${{ steps.release.outputs.tag }}
run: |
set -uo pipefail
# No --cleanup-tag: an unpublished draft never created a git tag.
echo "Run failed before publish; discarding draft $TAG"
gh release delete "$TAG" --repo "$HOURLY_REPO" --yes ||
echo "::warning::Could not discard draft $TAG; remove it manually."
- name: Prune old hourly releases
if: steps.freshness.outputs.should_build == 'true'
shell: bash
env:
GH_TOKEN: ${{ steps.app_token.outputs.token }}
run: |
set -euo pipefail
# Why: --cleanup-tag so pruning does not leave orphan tags behind that
# keep showing up in tag lists with no release or assets attached.
# Drafts are excluded so retention counts shipped builds only; a stale
# draft is handled by the failure path, not by the retention window.
stale="$(gh release list --repo "$HOURLY_REPO" --limit 200 --json tagName,createdAt,isDraft \
--jq "map(select(.isDraft | not)) | sort_by(.createdAt) | reverse | .[${HOURLY_RETAIN_COUNT}:] | .[].tagName")"
if [[ -z "$stale" ]]; then
echo "Nothing to prune; at or under $HOURLY_RETAIN_COUNT retained builds."
exit 0
fi
while read -r tag; do
[[ -n "$tag" ]] || continue
echo "Pruning $tag"
gh release delete "$tag" --repo "$HOURLY_REPO" --yes --cleanup-tag || \
echo "::warning::Could not prune $tag"
done <<<"$stale"