Works fine for me on master. Please try again with the next available snapshot.

@swift-ci create

Comment by Zhu Shengqi (JIRA)

The code still failed to compile with latest snapshot (January 24, 2020). I have provided the Xcode version and the snapshot version. Anyone want to fix the issue should try to reproduce it first, not just saying "works fine for me".

Sorry, I did try it yesterday and it worked for me. See:


Let me pull from master again and retry.

Comment by Zhu Shengqi (JIRA)

You are not using the same environment as I do. I use Xcode 11.3.1 (11C504) with development snapshot, while you are using terminal with swift git repo.

I don’t think Xcode is going to make any difference given this is a Swift error, but I will verify with Xcode as well. There are two snapshots, a “trunk” snapshot and a 5.2 development snapshot. They are different, so could you confirm which one you’re using exactly?

Comment by Zhu Shengqi (JIRA)

I use snapshot under the "Trunk development (master)" category.

Comment by Zhu Shengqi (JIRA)

I just download the "Swift 5.2 development" snapshot and the code compiles with it. So either the code in "Trunk development (master)" introduce a new bug, or a fix is committed in "Swift 5.2 development" but not merged back to "Trunk development (master)" yet. Can someone verify which is the case?

Comment by Zhu Shengqi (JIRA)

And if master branch mentioned above corresponds to the "Trunk development (master)", then there may be some problem when the CI create the snapshots.

Okay, I can reproduce it with the latest trunk snapshot and on the master branch I just pulled from. This is strange, because it compiled successfully before (as I showed earlier). So I think something has regressed since then.

Looking into this a bit more - this PR added a call to performSyntacticExprDiagnostics during builder transform, which is triggering this diagnostic. It seems like we're running the check a bit early - the closure does not have a type set, which means we're unable to determine if the closure is marked noescape or not.

This PR should fix the issue: #29649

@hborla, since PR was merged this should be resolved?

@LucianoPAlmeida oops, yes, thank you!

The issue has been fixed on master. Could you please verify using a recent master snapshot? Thank you!