CI (Continous Integration) / run Gradle tests for each PR?

[mccartney]

Would the project benefit if ./gradlew test was run for all raised PRs?

As far as I can tell there's nothing like this set up.

[FloEdelmann]

@westnordost doesn't like that. See #3649 (comment), #3760, #1917.

[FloEdelmann]

I can't currently find the exact quote, but he said that it should be as easy as possible to contribute and failing (for whatever reason) tests might be scaring in a PR.

ktlint for current master reports a lot of warnings, esp. deprecation-based. So every PR would have a lot of them (regardless if it introduced any or not).

That's not true; it's currently passing without any warnings (see latest workflow run). But I agree, there are often lint issues that are not that important to fix in a PR; successful compilation and passing tests would be more important.

[mccartney]

I can't currently find the exact quote, but he said that it should be as easy as possible to contribute and failing tests (for whatever reason) might be scaring in a PR.

OK, gotcha.

[mccartney]

That's not true; it's currently passing without any warnings

Sorry. You're right. I was confused by the output of ./gradlew lint run locally. It printed a lot of warnings, but looking at them now maybe they are not from the linter itself, but just the compiler.

[FloEdelmann]

Note that ./gradlew lint and ktlint are not the same thing.

[peternewman]

I can't currently find the exact quote, but he said that it should be as easy as possible to contribute and failing (for whatever reason) tests might be scaring in a PR.

Although we could always make the tests passing be optional rather than mandatory which might make them less scary (if less effective).

For anyone else who finds this, in #3965 I added the ability to run them easily manually via an action at least.