Fix Signatures.kt:buildFunctionSignature

[VitalyVPinchuk]

For #4887 ('buildFunctionSignature' fails when function body has KDoc)

[VitalyVPinchuk]

:detekt-api:test fails, but locally it passes alright. Why is that?

[BraisGabin]

You can find information about that here: #4874. Any feedback there is more than welcome.

[BraisGabin]

Thanks!

[BraisGabin]

This is a fix but it's also kind of "breaking". It will force to regenerate lots of baselines. I'm ok with that but just a heads up if some maintainer wants to pospone he merge to 1.22.0. cc @cortinico

[cortinico]

This is a fix but it's also kind of "breaking". It will force to regenerate lots of baselines. I'm ok with that but just a heads up if some maintainer wants to pospone he merge to 1.22.0

I think we can merge this, but would be great to include a paragraph in the 'migration' section of the changelog.

[BraisGabin]

I think that something like this could do the trick on the changelog:

We updated the baseline creation. Now the kdoc is not part of the function signature. This could create that already suppressed issues in your baseline reappear. To fix that just recreate your baseline.

[VitalyVPinchuk]

I think that something like this could do the trick on the changelog:

We updated the baseline creation. Now the kdoc is not part of the function signature. This could create that already suppressed issues in your baseline reappear. To fix that just recreate your baseline.

I think KDoc comments were trimmed before that too.

element.docComment?.let {
    methodStart = it.endOffset - it.startOffset
}

[cortinico]

I think that something like this could do the trick on the changelog:
We updated the baseline creation. Now the kdoc is not part of the function signature. This could create that already suppressed issues in your baseline reappear. To fix that just recreate your baseline.

I think KDoc comments were trimmed before that too.

element.docComment?.let {
    methodStart = it.endOffset - it.startOffset
}

Can we verify this. If we require users to regenerate baselines because of trimming of KDocs, that should be reported clearly in the changelog 👍

[VitalyVPinchuk]

@cortinico No, I was wrong!
Previous revision doesn't pass current tests because of leaving comments in the signature!

org.opentest4j.AssertionFailedError: 
expected: "Test.kt$fun data()"
 but was: "Test.kt$/* comments */ fun data()"

Consider the issue #4767

[cortinico]

Previous revision doesn't pass current tests because of leaving comments in the signature!

Still this doesn't answer my original question: do we need to ask users to regenerate baselines?

[VitalyVPinchuk]

Still this doesn't answer my original question: do we need to ask users to regenerate baselines?

Yes, we do.