release.yml: run frontend + Rust tests before building a release

[DianaSensei]

Summary

tests.yml (frontend vitest) and coverage.yml (frontend vitest + Rust via cargo llvm-cov) only trigger on pull_request/push: main — neither ever runs on a tag push, and workflow_dispatch here accepts any ref. release.yml's build job went straight into tauri build, which only typechecks as a side effect of npm run build's tsc && vite build — npm test and cargo test never ran anywhere in this pipeline. A release tag pushed on a commit whose CI run was stale, skipped, or simply never existed for that ref would build and publish installers with nothing having actually verified the test suites pass.

This is the same class of gap just found and fixed across developer-desktop-util-plugin's four release.yml files (that repo's release build never ran cargo test either).

Changes

Added a test job to release.yml:

• npx tsc --noEmit
• npm test (vitest)
• cargo test --manifest-path src-tauri/Cargo.toml — using the same externalBin placeholder-binary technique coverage.yml's rust job already relies on, since tauri-build's build.rs checks those paths exist even for a plain cargo test.

build now depends on test via needs: test, so it's gated for both the tag-push trigger and workflow_dispatch.

Test plan

• npx tsc --noEmit — clean
• cargo test --manifest-path src-tauri/Cargo.toml — 115/115 passing (106 unit + 9 integration)
• YAML validated (python3 -c "import yaml; yaml.safe_load(...)")

🤖 Generated with Claude Code

https://claude.ai/code/session_01BCypCuViyQDKGWxWspXKs2

---

Generated by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 46.12%. Comparing base (0e97e63) to head (aad9d59).
⚠️ Report is 14 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #156   +/-   ##
=======================================
  Coverage   46.12%   46.12%           
=======================================
  Files         313      313           
  Lines       21663    21663           
  Branches     5294     5294           
=======================================
  Hits         9992     9992           
  Misses      10485    10485           
  Partials     1186     1186           

Flag	Coverage Δ
frontend	40.73% <ø> (ø)
rust	69.90% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[pullfrog]

Caution

The new test job does not create ../dist, but cargo test compiles the devtool bin target where tauri::generate_context!() panics at compile time if frontendDist is missing. On a fresh runner (no dist), this job fails and, via needs: test, blocks every release. See the inline comment.

Reviewed changes

• New test job — runs npx tsc --noEmit, npm test, and cargo test on ubuntu-22.04, restoring the coverage Rust cache read-only and creating placeholder externalBin files.
• build gated by needs: test — tag-push and workflow_dispatch runs now require the test job to pass first.

The intent is sound and closes a real gap: tests.yml is PR-only and coverage.yml runs only on PR/push: main, so a tag built from a commit with no prior CI run was never tested. The externalBin placeholder technique is copied correctly from coverage.yml.

The one blocker is that the job is missing the frontend build step coverage.yml needs for the same cargo invocation.

---

Important

Pullfrog covered this run's model usage. DeepSeek Flash is fast and cheap — expect lighter work than a frontier model. This model allowance is temporary and separate from your Pullfrog plan. Connect a model-provider subscription or API key →

  ^{｜ Fix all ➔ ｜ Fix 👍s ➔ ｜ View workflow run ｜ Using DeepSeek Flash (model usage covered by Pullfrog) ｜ 𝕏}

[pullfrog]

ℹ️ No new issues in this delta. The release.yml test job is unchanged, so the prior review's blocking finding there still stands; the only new note is a doc nitpick.

Reviewed changes

• Frontend coverage moved into tests.yml — the frontend job now runs npm run test:coverage and uploads coverage/lcov.info to Codecov under the frontend flag, replacing the plain npm test it ran before.
• coverage.yml's frontend job scoped to push: main — added if: github.event_name != 'pull_request', removing the byte-for-byte duplicate npm ci + vitest run on PRs; the job stays on main to keep the Codecov baseline fresh.

The split is sound: tests.yml is PR-only, so main still gets exactly one frontend upload and PRs get exactly one, with no gap. rust is untouched and still runs on both triggers, and npm run test:coverage resolves to the same suite (vitest run --coverage) with no thresholds, so the required Typecheck and unit tests check is unaffected.

ℹ️ Nitpicks

• codecov.yml:2-3 still says both flags are "uploaded by .github/workflows/coverage.yml's two jobs". After this change the frontend flag is uploaded by tests.yml on PRs and by coverage.yml on push: main, so that comment is now stale.

---

Important

Pullfrog covered this run's model usage. DeepSeek Flash is fast and cheap — expect lighter work than a frontier model. This model allowance is temporary and separate from your Pullfrog plan. Connect a model-provider subscription or API key →

  ^{｜ Fix it ➔ ｜ View workflow run ｜ Using DeepSeek Flash (model usage covered by Pullfrog) ｜ 𝕏}

[pullfrog]

✅ No new issues found.

Reviewed changes

• Replaced the release test job's npx tsc --noEmit step with npm run build (tsc && vite build), producing the ../dist that tauri::generate_context!() reads at macro-expansion before cargo test compiles the devtool bin target.

This resolves the prior blocking finding: the test job now mirrors coverage.yml's rust job prerequisites (setup with the shared coverage cache key → npm run build → placeholder externalBins → cargo test) and no longer fails on a fresh runner. The previous thread has been replied to and resolved.

---

Important

Pullfrog covered this run's model usage. DeepSeek Flash is fast and cheap — expect lighter work than a frontier model. This model allowance is temporary and separate from your Pullfrog plan. Connect a model-provider subscription or API key →

  ^{｜ View workflow run ｜ Using deepseek-v4.1-flash (model usage covered by Pullfrog) ｜ 𝕏}

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud