False positive clang warning with braced initializer

Question

False positive clang warning with braced initializer

Closed this issue 4 years ago · 6 comments

On OS X, clang throws an erroneous warning for this correct code:

https://github.com/RoverRobotics-forks/rmw_cyclonedds/blob/a850bec48226b832db469c48c9da32e170bffbf0/rmw_cyclonedds_cpp/src/Serialization.cpp#L291

https://ci.ros2.org/job/ci_osx/7277/warnings11Result/new/

Answer 1 · 2019-12-12T21:44:02.000Z

Reference to the ticket where this came up: ros2/rmw_cyclonedds#80 (comment)

@rotu What is the goal / acceptance criteria of this ticket? I would guess clang would need to fix this false positive warning. If that is the case please close the ticket since that is not in the scope of this repository. Maybe consider create a ticket upstream for clang?

Answer 2 · 2019-12-12T23:15:45.000Z

The goal is to modify CI to turn off or whitelist the warning. This is already fixed in more recent Clang versions: https://bugs.llvm.org/show_bug.cgi?id=21629

Answer 3 · 2019-12-12T23:50:38.000Z

Please consider to provide a pull request for the proposal. If this had been addressed upstream and also can be worked around by code I don't think we will be spending resources in this.

Answer 4 · 2019-12-13T01:39:35.000Z

I don’t know if I understand CI enough to do the PR. This also might be related to using the obsolete warnings plugin (#315). I don’t know why a build warning should cause CI to fail when we have dedicated linters for those checks.

Answer 5 · 2019-12-13T04:33:09.000Z

A build warning does not fail the Jenkins build - it only marks it as unstable (in order to distinguish it from all green builds). Only the GitHub status API doesn't support a tri state and considers unstable the same as a failed build.

Answer 6 · 2021-02-14T16:25:54.000Z

I'm pretty sure this hasn't been a problem in a while (we have had green macOS jobs for a while). So I'm closing this out, but feel free to reopen if you still see a problem here.