@swift-ci create

In ValueDecl::getOverloadSignature, we’re canonicalizing the result of calling mapSignatureFunctionType. It also looks like within mapSignatureFunctionType we use getAs<TupleType> which will walk past the ParenType. If we then reform the type we end up dropping the ParenType:

      if (anyChanged) {
        argTy = TupleType::get(elements, ctx);

Same issue as SR-5112.

Sorry for the extra traffic, I just wanted to make sure all the various test cases that expose the same bug are at least linked in the bug tracker with proper links. Jens, if the devs sometimes forget to set the duplicate in the system and just comment about that, you can do it for them. Instead of `Close Issue`, use `Resolve Issue` first. From `Resolution` drop down select `Duplicate`, then from `Linked Issues` drop down select “duplicates” and fill the master bug in the `Issue` field. Now you can finally `Resolve`. After all this is done, `Close Issue`. I know proper process hygiene isn’t much fun in JIRA, but it helps to keep thing in order. This is Swift’s institutional memory.

Thank you Pavol, much appreciated, will try to remember that next time!

Just to make sure I'm not doing this wrong too:

When one of my issues has been marked "resolved" by the devs, after fixing the issue, am I correct in that I should "close" it after verifying that I can no longer reproduce it?

From memory... (Can't find other docs than this forum post) I think you should close it after you verify it solved the problem for you.

It's a bit unclear, but I think if you should wait for the fix to ship in a release version.
(Probably depends on the version you're reporting from...)