Is your feature request related to a problem? Please describe.
The test suite has grown large enough that running it locally is becoming expensive during development.
A full run currently takes around 12 minutes on my machine:
Tests: 1080 passed (11968 assertions)
Duration: 738.14s
This makes running the full suite frequently while working on a PR impractical and slows down the feedback loop.
Describe the solution you'd like
I think we should evaluate migrating the test runner to Pest 5, mainly to take advantage of its new Test Impact Analysis (TIA) support.
TIA keeps track of which tests depend on which parts of the application and, after the initial baseline run, can rerun only the tests affected by the current changes.
This could make the normal development workflow much faster without reducing the coverage of the full test suite in CI.
Since OGameX is already on PHP 8.5 and PHPUnit 13, the project should already meet Pest 5's main runtime requirements.
I would suggest:
- Evaluate running the existing suite through Pest 5 without immediately rewriting all tests.
- Benchmark the current full suite against Pest.
- Test
--tia during normal local development.
- Keep a complete test-suite run in CI to ensure nothing is missed.
- If the results are useful, gradually migrate tests to Pest syntax rather than doing a large rewrite.
Describe alternatives you've considered
We could optimize individual slow tests or increase parallelization with the existing PHPUnit/ParaTest setup.
Those are still worth doing, especially for improving the actual full-suite runtime, but they do not solve the common development case where a small change causes us to repeatedly run a large suite.
Pest 5/TIA could complement those optimizations rather than replace them.
Additional context
Current local result:
Tests: 1080 passed (11968 assertions)
Duration: 738.14s
Screenshot attached.
The goal here is not necessarily to migrate the whole test suite to Pest syntax. The first step would be to benchmark whether Pest 5/TIA gives us a meaningful improvement in developer feedback time.
Locally:
Is your feature request related to a problem? Please describe.
The test suite has grown large enough that running it locally is becoming expensive during development.
A full run currently takes around 12 minutes on my machine:
This makes running the full suite frequently while working on a PR impractical and slows down the feedback loop.
Describe the solution you'd like
I think we should evaluate migrating the test runner to Pest 5, mainly to take advantage of its new Test Impact Analysis (TIA) support.
TIA keeps track of which tests depend on which parts of the application and, after the initial baseline run, can rerun only the tests affected by the current changes.
This could make the normal development workflow much faster without reducing the coverage of the full test suite in CI.
Since OGameX is already on PHP 8.5 and PHPUnit 13, the project should already meet Pest 5's main runtime requirements.
I would suggest:
--tiaduring normal local development.Describe alternatives you've considered
We could optimize individual slow tests or increase parallelization with the existing PHPUnit/ParaTest setup.
Those are still worth doing, especially for improving the actual full-suite runtime, but they do not solve the common development case where a small change causes us to repeatedly run a large suite.
Pest 5/TIA could complement those optimizations rather than replace them.
Additional context
Current local result:
Screenshot attached.
The goal here is not necessarily to migrate the whole test suite to Pest syntax. The first step would be to benchmark whether Pest 5/TIA gives us a meaningful improvement in developer feedback time.
Locally:
