CI design for the test suite #8

Activity

1. dtinth-claw

   added

   wayfinder:grillingWayfinder grilling ticketWayfinder grilling ticket

   on Jul 22, 2026
2. dtinth-claw

   added a parent issue on Jul 22, 2026

   • Testing harness #3
3. dtinth-claw

   marked this as blocked by 2 issues on Jul 22, 2026

   • Prove the harness: QtTest seed suite builds and passes in the fork #4
   • Upstream contribution requirements for a test/CI PR #5
4. dtinth-claw commented on Jul 22, 2026

   

   dtinth-clawBot

   on Jul 22, 2026 – with dtinth-claw

   Author

   More actions

   🔒 Claimed by agent (orchestrator dispatch, 2026-07-22) — ci-shepherd agent wiring up the workflow on testing-harness and shepherding it to green on Linux/macOS/Windows.
5. dtinth-claw commented on Jul 22, 2026

   

   dtinth-clawBot

   on Jul 22, 2026 – with dtinth-claw

   Author

   More actions

   Resolution

   CI is wired and verified green on all three platforms: run 29926662716 — every job's log contains Totals: 41 passed, 0 failed, 0 skipped, 0 blacklisted.

   Design as landed (.github/workflows/unit-tests.yml on testing-harness):

   • Single "Unit Tests" job, 4-config matrix, fail-fast: false:

     config	Qt	how	duration
     ubuntu-22.04	5.15 (apt)	qmake + make	26s
     ubuntu-24.04	6.x (apt)	qmake6 + make	36s
     macos-15	6.10.2 (install-qt-action)	+ Xcode 16.3.0 pin (mirrors autobuild)	49s
     windows-latest	6.10.2 MSVC	qmake + nmake	66s

   • permissions: contents: read, SHA-pinned third-party actions, paths: filters — house style per the #5 research.
   • Hard execution gate on every platform: output must contain Totals: 41 passed (grep on Unix, findstr on Windows) — a job can no longer pass on exit code alone.

   Findings worth keeping:

   1. The Qt 5.15 job compiled and passed with zero source changes — the suite is genuinely Qt5-clean; the only platform fix ever needed was NOMINMAX in test.pro on Windows.
   2. macos-latest (Xcode 26) cannot compile Qt 6.x headers (qyieldcpu.h __yield); the fix is upstream autobuild's own convention (macos-15 + Xcode 16.3.0, same pinned action SHA) — an upstream PR should reuse it.
   3. Windows oddity: the test exe's stdout was silently 0 bytes under every capture method (cmd redirect, pwsh Tee/redirect). Sidestepped robustly with QTest's own -o test-output.txt,txt file writer; exact culprit unidentified (suspect console-handle inheritance in the runner's nested cmd tree). Noted in case a future Windows job needs another console app's stdout.
   4. Still to reshape before upstreaming (flagged in a comment in the workflow): triggers are currently workflow_dispatch + push-to-testing-harness for iteration; the upstream version should use push/pull_request on main with the same paths: filters.

   Feeds Scope the first upstream PR, which is now the only open ticket.
6. dtinth-claw

   closed this as completedon Jul 22, 2026