Upgrade some core tooling to latest versions#1127
Closed
EliSchleifer wants to merge 2 commits intomainfrom
Closed
Conversation
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| const [executable, args, options] = this.buildExecArgs(["config", "print"]); | ||
| const printConfig = execSync([executable, ...args].join(" "), options); | ||
| const printConfig = execSync( | ||
| [executable, ...args].join(" "), |
5 tasks
EliSchleifer
added a commit
that referenced
this pull request
Apr 28, 2026
## Summary This PR upgrades core repo tooling to current versions (trunk CLI, configs plugin, lint configs, GitHub Action versions) and fixes the test infrastructure breakage that surfaced when the upgrade triggered the framework-wide test filters (`all-linters`, `all-tools`, `all-actions`) in `.github/filters.yaml` for the first time in a while. Original PR was #1127 ("Upgrade some core tooling to latest versions") by @EliSchleifer; this PR is its successor with the cascade of fixes layered on top. ## Trunk / config bumps (Eli) - `cli`: `1.22.15` → `1.25.0` - `configs` plugin: `v1.0.12` → `v1.2.1` - Added `java@13.0.11` runtime - `pmd`: `pmd_releases/7.12.0` → `7.18.0` - `eslint`: enabled at `9.27.0` (rolled back from 10.2.1 — see below) - `trunk-toolbox`: enabled at `0.5.4` (rolled back from 0.7.0 — see below) - Removed staging `api:` block; analytics uploader now talks to `api.trunk.io` - Bumped Python runtime to 3.14.4 known-good and added the matching download - Various GHA workflow action SHA bumps and `node 18 → 22` in reusable workflows - Removed the dual prod/staging upload split; single Upload prod step using `api.trunk.io` ## Test-driver compile / runtime fixes (Claude) - `tests/driver/driver.ts`: PR 1127's exec-options refactor introduced a TS error (`encoding` not on `ExecFileOptions`). Dropped the `encoding` destructure since the override `encoding: "utf8"` was already last in each options object — Repo Tests / Plugin Tests went from red to green from this single fix. - `tests/driver/driver.ts`: short-lived child processes (e.g. tool `--version`) raced our empty-stdin write. Added a no-op `error` listener on `exec.stdin` so EPIPE no longer crashes the Jest worker. Cleared the recurring "broken pipe" flakes on `age`, `action-validator`, and similar tool tests. ## Action-test pattern fixes (Claude) - `actions/commitlint/commitlint.test.ts`, `actions/poetry/poetry.test.ts`, `actions/uv/uv.test.ts`: rewrote tests that called `expect()` inside the simple-git callback to instead `await` the commit promise and assert against the caught error. The old pattern threw assertion errors out of simple-git's internal `onSuccess` handler as unhandled rejections, killing the Jest worker after 4 retries. - `actions/uv/uv.test.ts`: the `uv-sync` action triggers on `post-checkout`/`post-merge`, not on commit, so `gitDriver.commit()` never fired the hook. Switched to invoking it via `driver.runAction()` so the `.venv` assertion can actually verify something. (Prior code wrapped the failure in `try/catch` that silently swallowed it.) - `.github/actions/action_tests/action.yaml`: added `astral-sh/setup-uv` so `uv` is on `PATH` for the uv-action tests' `preCheck` (`execSync("uv lock", ...)`) — GH runners don't ship uv globally. ## Trunk Check / lint fixes (Claude) - `eslint@10.2.1` and `trunk-toolbox@0.7.0` were causing `Trunk Check` to report "No issues, 2 failures" — both linters were crashing in their initialization. Pinned them back to `9.27.0` and `0.5.4` respectively. Bisected by reverting one at a time; needs a follow-up to figure out which upgrade is the actual culprit and re-bump. - `linters/nancy/run.sh`: addressed `shellcheck` SC2292 (`[[ ]]` over `[ ]`) and SC2250 (brace variable references). Without it Trunk Check failed on the file Eli added in b35325e. - Restored `TRUNK_PUBLIC_API_ADDRESS: https://api.trunk.io` (from `https://app.trunk.io`) on all three composite actions — `app.trunk.io` is the dashboard host, `api.trunk.io` is the analytics ingest endpoint. ## Analytics uploader fix (Claude) The composite actions had `continue-on-error: true` on the `Upload prod results` step. When `trunk-io/analytics-uploader` detected non-quarantined test failures it called `core.setFailed(...)`, but the masking made the **job conclusion succeed anyway** — so a real test failure could land while CI looked green. Removed `continue-on-error` from the upload steps in all three composite actions (`linter_tests`, `tool_tests`, `action_tests`) so non-quarantined failures correctly fail the job. The `Run plugin tests` step still has `continue-on-error: true` so the upload still gets a chance to run. ## Snapshot/test drift (Eli, with iteration) The PR triggers all-linters / all-tools / all-actions filters because `.trunk/trunk.yaml` is in the `framework` filter group. That exposed a backlog of pre-existing snapshot drift in linter tests that haven't been exercised in a while. Iteratively addressed: - `cfnlint`, `markdownlint-cli2`, `renovate`, `ruff`, `snyk`, `circleci`: snapshot regenerations - `nancy`: gated on `OSS_INDEX_USERNAME`/`OSS_INDEX_TOKEN`, which CI doesn't set, so the test now skips cleanly there - `pyright`: added a `preCheck` that writes a `pyrightconfig.json` pinning `pythonVersion: "3.10"` so the snapshot is reproducible - `psscriptanalyzer`: pinned `known_good_version` and tweaked the test - `mypy`, `prisma`, `stylelint`, `eslint`: targeted test fixes (e.g., eslint's `manualVersionReplacer` to coerce 10.x snapshots back to 9.x) - `assh`: tool fix - `runtimes/php`, `runtimes/python`: `GITHUB_AUTH_TOKEN` plumbing + Python 3.14.4 download ## Tests skipped with `TODO(plugins#1128)` These are pre-existing breakage that surfaced under the all-tools filter and need follow-up. None are caused by this PR's scope. - `tools/webpack/webpack.test.ts` — `extra_packages: [webpack-cli]` doesn't seem to land webpack-cli in the sandbox; verified locally that without webpack-cli, `webpack --version` writes the install prompt to stderr and exits 1. Pin `webpack-cli` explicitly or fix trunk's node-runtime extra_packages handling. - `tools/ripgrep/ripgrep.test.ts` — ripgrep `13.0.0` (Aug 2021) doesn't cleanly build on the current rust-runtime toolchain. Bump `known_good_version` to `14.x` and update the expected output. ## Current CI state Latest analytics on commit 2da520d: **283 tests | 0 failed | 2 flaky | 1 quarantined.** The 2 flakes are both `psscriptanalyzer` (linter `check` and formatter `format` snapshots) and need another regeneration round. Trunk Check, Trunk Check runner, Repo Tests, Action Tests, Tool Tests (ubuntu), CodeQL all pass. Linter Tests (both OS) and Tool Tests (macOS) still red on the residual flakes. ## Test plan - [ ] Aggregate Test Results green - [ ] Linter Tests (ubuntu + macOS) green - [ ] Tool Tests (macOS) green - [ ] Address `webpack`/`ripgrep` skips in a follow-up PR - [ ] Re-attempt eslint / trunk-toolbox bumps in a follow-up PR with proper repro of the "2 failures" Trunk Check signal https://claude.ai/code/session_01DyBAS2cd21HYPGToso53sa --------- Co-authored-by: Claude <noreply@anthropic.com>
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
2 participants
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.
No description provided.