Fix -Wdocumentation warning. NFCI.

[JDevlieghere]

llvm-svn: 372418
(cherry picked from commit 60a6f31)

[JDevlieghere]

@swift-ci please test

[cyndyishida]

Since the update-checkout script doesn't call this dated branch anywhere, how do you check if the PR test failure is because of a dirty baseline, check that swift/master is a reflection and check those CI jobs?

[JDevlieghere]

I don't think PR testing is using update-checkout at all. @shahmishal knows the details.

The linux failure is definitely because of a dirty baseline, which is why Linux testing in temporarily not required. When that's fixed you should always (in theory) be able to rely on a clean baseline. However, in practice, flakiness or configuration errors mean that it is possible to end up with a dirty baseline. When I'm suspicious I usually check a few subsequent PR-test runs and see if there's a pattern.

[shahmishal]

I don't think PR testing is using update-checkout at all. @shahmishal knows the details.

The linux failure is definitely because of a dirty baseline, which is why Linux testing in temporarily not required. When that's fixed you should always (in theory) be able to rely on a clean baseline. However, in practice, flakiness or configuration errors mean that it is possible to end up with a dirty baseline. When I'm suspicious I usually check a few subsequent PR-test runs and see if there's a pattern.

it's not using update-checkout because for apple/* branches don't get tested for swift changes.

Job it self is checking out the PR branch git rev-parse refs/remotes/origin/origin/pr/213/merge^{commit} # timeout=10

[shahmishal]

#1 (comment) I was seeing this failure yesterday when I created Linux PR testing for LLVM Project. (cc: @hyp)

[shahmishal]

@swift-ci clean test Linux