It seems that the execution time is slightly slower now compared to before, though it’s hard to tell exactly, because there is significant variance between passes.
In the reports linked above, you see an execution time of ~12s (after) vs. ~9s (before). The bootstrapping procedure of the CI job (installing dependencies, etc.) is usually 60–90 seconds, however, so I’m not sure we should be worried too much at this point. I also think we should have more leverage now to speed up the tests through parallelizing.
test clauses are usually run in parallel by default in Playwright, though it seems we opted out of test parallelism in our Playwright config. I’m not sure why we did that, but we could now consider changing it back. (If at all, we should do that separately, though.)
This PR is a non-functional refactoring. It only introduces the grouping + group descriptions, without any other modifications (except code style).
Tagged @cghague for review, since we just ran into this topic recently, so I thought it was fitting. I’d take care of adjusting the Pro-specific tests once we have merged this over.
As demonstrated in https://github.com/tiny-pilot/tinypilot/pull/1685, we can make use of Playwright’s grouping feature to improve our e2e test structure:
test
clauses. These blocks are grouped inside an enclosingtest.describe
.test
clauses are usually run in parallel by default in Playwright, though it seems we opted out of test parallelism in our Playwright config. I’m not sure why we did that, but we could now consider changing it back. (If at all, we should do that separately, though.)This PR is a non-functional refactoring. It only introduces the grouping + group descriptions, without any other modifications (except code style).
Tagged @cghague for review, since we just ran into this topic recently, so I thought it was fitting. I’d take care of adjusting the Pro-specific tests once we have merged this over.