User tests: Successful: Unsuccessful:
This Pull Request improves CI workflow stability and fixes multiple configuration issues.
This PR fixes multiple issues in .github/workflows/ci.yml related to CI reliability, caching behavior, artifact uploads, and workflow safety.
The sleep 3 command inside a YAML literal block (|) used inconsistent indentation (9 spaces instead of 10), which could lead to readability issues and potential parsing inconsistencies.
The tests-integration-windows job referenced:
steps.cache-php-windows.outputs.cache-hit
However, the cache restore step was missing:
id: cache-php-windows
This caused Composer dependencies to reinstall on every run even when cache restoration succeeded.
The tests-system matrix job previously used a static artifact name:
system-test-output
Since actions/upload-artifact@v4 requires unique artifact names, parallel matrix executions (6 DB configs × 2 browsers) caused upload conflicts.
Artifact names now include matrix variables:
system-test-output-${{ matrix.config.test_group }}-${{ matrix.browser }}
Also fixes typo: Archive test results results → Archive test results
No jobs defined timeout-minutes, defaulting to GitHub's 360-minute limit.
Reasonable timeout limits were added to prevent runaway builds and unnecessary runner consumption:
| Job | Timeout |
|---|---|
composer |
10 min |
npm |
15 min |
code-style-php |
10 min |
code-style-js-css |
10 min |
phpstan |
15 min |
tests-unit |
15 min |
tests-unit-windows |
15 min |
tests-integration |
30 min |
tests-integration-windows |
30 min |
tests-system-prepare |
15 min |
tests-system |
60 min |
typos |
5 min |
Please select:
Documentation link for guide.joomla.org:
No documentation changes for guide.joomla.org needed
Pull Request link for manual.joomla.org:
No documentation changes for manual.joomla.org needed
| Status | New | ⇒ | Pending |
| Category | ⇒ | Repository |
| Title |
|
||||||
Thanks for your work. Unfortunately this pr does not improve the workflow so I'm closing it it to save valuable resources for maintainers.
| Status | Pending | ⇒ | Closed |
| Closed_Date | 0000-00-00 00:00:00 | ⇒ | 2026-02-27 16:59:57 |
| Closed_By | ⇒ | laoneo | |
| Labels |
Added:
PR-5.4-dev
|
||
@D3S-Gaurav This PR here obviously also contains the changes from your other PR #46926 , and the description of this PR tells so. So it seems this PR here shall replace the other one. Why don't you close the other one? Having multiple concurrent PRs open causes additional work for maintainers. Every author is able to close their own PRs.