k6 specifically runs some tests that usually fail and arguably a lot of stuff that aren't implemented at all.
This has helped a lot over the years to catch changes that break tests or fix problems due to dependancy updates. It also significantly redueces the maintaince burden as instead of the developer needing to go through a list of hardcoded test that fail - the list is generated when you run the test.
This should probably also go with removing any skipping of tests that aren't just "We skip this as this is not part of the specificaiton yet".
Downsides:
bigger test times - this definitely is a problem, but if you are working on a feature it is way easier to see that the tests you are having are not failing, than when you have to also find what test you need to run.
less compatible with goja - this can be fixed by us backporting those changes to goja as well.
k6 specifically runs some tests that usually fail and arguably a lot of stuff that aren't implemented at all.
This has helped a lot over the years to catch changes that break tests or fix problems due to dependancy updates. It also significantly redueces the maintaince burden as instead of the developer needing to go through a list of hardcoded test that fail - the list is generated when you run the test.
This should probably also go with removing any skipping of tests that aren't just "We skip this as this is not part of the specificaiton yet".
Downsides: