fix(ci): rewire typecheck/test workflows for local/amicode default branch - #2
Merged
Merged
Conversation
jack-champagne
force-pushed
the
ci/rewire-workflows-for-fork
branch
from
July 8, 2026 05:43
dcf2cb3 to
b8b15f9
Compare
jack-champagne
force-pushed
the
ci/rewire-workflows-for-fork
branch
from
July 28, 2026 03:06
024a1fc to
0fbbe4b
Compare
This was referenced Jul 28, 2026
…anch These workflows were inherited from upstream sst/opencode, which uses `dev` as its default branch. The fork's default is `local/amicode`, so the `dev` branch gates never matched: - typecheck.yml only ran on `pull_request: branches: [dev]`, so it never ran on any fork PR (which target local/amicode). Typechecking was effectively disabled. Drop the PR branch filter (run on all PRs, like test.yml) and point the push trigger at local/amicode. - test.yml's push trigger and concurrency guard referenced `dev`; point them at local/amicode so post-merge runs fire and default-branch runs aren't cancelled. - Pin the e2e job to Node 24.15 (upstream's workaround for a Playwright 1.59 Chromium-extraction hang on 24.16); the fork was on floating "24". Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The fork inherited `blacksmith-4vcpu-*` runner labels from upstream, but the org has zero Blacksmith runners registered, so every job queued forever and no test/typecheck run ever executed here. Switch to GitHub-hosted ubuntu-latest / windows-latest so CI actually runs. Suite verified green locally on Linux (opencode 2980, ui 95, app 376, httpapi 172, typecheck 23/23). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The suite's tight timing budgets were tuned for upstream's Blacksmith runners; on GitHub-hosted runners two tests overshoot marginally: - run-process.test.ts: unknown-model prompt-exit budget 15s -> 18s (was 15264ms on CI). Bump both the harness timeoutMs and the assertion so the "exits on its own before the kill deadline" regression invariant still holds (a real hang fails at the 18s deadline; outer timeout 30s). - app e2e: APP_READY_TIMEOUT 30s -> 45s and CI workers 5 -> 2. The heaviest smoke test (seeded timeline) timed out reaching app-ready under 5-worker contention on 4 vCPUs; fewer workers + more headroom. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Events carry a raw Location.Ref ({directory, workspaceID?}) but the v2
wire contract is a resolved Location.Info — server.connected was built
resolved while every subsequent event streamed unresolved, so schema
decoding (location.project) failed for all consumers. The stream filter
already pins events to the connection's location; map them to the same
resolved info.
Also adapt the timeline smoke e2e to the amicode home: a single seeded
project is auto-focused and its row deliberately not rendered
(HomeProjectList shows rows only when >1 projects), so wait on the
home-cards strip + focused-project content instead of clicking a row.
linux finishes in ~15min; windows timed out at the 20min cap mid-suite with zero test failures.
jack-champagne
force-pushed
the
ci/rewire-workflows-for-fork
branch
from
July 28, 2026 20:29
0fbbe4b to
81dea5d
Compare
This was referenced Aug 10, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The
typecheckandtestworkflows were inherited from upstreamsst/opencode, which usesdevas its default branch. This fork's default islocal/amicode, so the hard-codeddevgates silently disabled coverage.Changes
typecheck.yml— was gated onpull_request: branches: [dev], so it never ran on any fork PR (all targetlocal/amicode). Typechecking was effectively off. Dropped the PR branch filter (now runs on all PRs, matchingtest.yml) and pointedpushatlocal/amicode.test.yml— pointed thepushtrigger and the concurrency guard atlocal/amicode(post-merge runs now fire; default-branch runs aren't cancel-in-progress'd). PR-level behavior is unchanged — it already ran on all PRs.24.15(upstream's workaround for the Playwright 1.59 Chromium-extraction hang on 24.16); the fork was on floating"24".Context
Companion to disabling the inherited upstream PR-hygiene bots (
pr-standards,pr-management,compliance-close,close-prs,close-issues), which crashed on the same nonexistentdevref. This PR is the "make the real tests run" half.🤖 Generated with Claude Code