A large portion of the test run in under 15 seconds, but there are a few larger tests:
- kernel tests (> 200 seconds)
- (maybe help tests - download figure)
- (maybe UI tests) (~ 20 seconds locally)
By moving the large tests into a separate CI stage, we could receive earlier feedback on failures.
Things to consider:
- How large would the overhead of running smalltalkCI twice be? How can we reuse the image?
- Apply the same solution to SimulationStudio as well
- Are there further outliers in the kernel tests that could be avoided/eliminated?
A large portion of the test run in under 15 seconds, but there are a few larger tests:
By moving the large tests into a separate CI stage, we could receive earlier feedback on failures.
Things to consider: