Created attachment 196274 [details] svn diff All tests pass with MPICH and OpenMPI. OpenMPI2 - some tests seem to hang, still because MPICH is the default, I think the port is ready. This PR makes https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=228814 obsolete.
Note that running tests *as root* is still causing a hang: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=226722
Created attachment 196287 [details] svn diff Forgot to bump SOVERSION for gcc8.
I applied the patch and build lang/opencoarrays 2.2.0 without problems in 11.2. I ran make test both as regular user and as root. A lot of tests failed but the system didn't hang up. Funny enough, building in 11.1{amd64,i386} in a jailed environment pass the tests when executed as root. It does not produce a hang either.
Created attachment 196338 [details] All tests pass with GCC8
Created attachment 196339 [details] But yes, many failures with GCC7
I'll report this upstream
A commit references this bug: Author: swills Date: Sun Aug 19 12:54:10 UTC 2018 New revision: 477570 URL: https://svnweb.freebsd.org/changeset/ports/477570 Log: lang/opencoarrays update to 2.2.0 PR: 230689 Submitted by: Anton Shterenlikht <as@cmplx.uk> (maintainer) Changes: head/lang/opencoarrays/Makefile head/lang/opencoarrays/distinfo
Committed, thanks!
Unfortunately it seems you applied the old (obsolete) patch. The correct patch is in attachment https://bz-attachments.freebsd.org/attachment.cgi?id=196287 See comment 2 Thanks!
(In reply to Steve Wills from comment #8) Why did we commit this patch? Shouldn't the problem with the tests be solved before proceeding with it? It still fails in my machine (11.2 amd64).
Comment on attachment 196287 [details] svn diff Correctly set maintainer approval
https://github.com/sourceryinstitute/OpenCoarrays/issues/581 upstream PR, for easy reference
I was under the mistaken impression the issue was an issue with the tests themselves. Should I go ahead and commit the rest of the change or wait for a complete fix?
I think the issue is indeed "with the tests themselves". I'm dealing with the upstream to investigate. I bet that after switching to GCC8, they haven't tested much with GCC7, hence the failures. I'd prefer to commit the rest of the patch now, as it is the correct patch.
A commit references this bug: Author: swills Date: Wed Aug 22 14:58:56 UTC 2018 New revision: 477793 URL: https://svnweb.freebsd.org/changeset/ports/477793 Log: lang/opencoarrays: Commit missed part of patch PR: 230689 Submitted by: Anton Shterenlikht <as@cmplx.uk> (maintainer) Reported by: Anton Shterenlikht <as@cmplx.uk> (maintainer) Pointyhat to: swills Changes: head/lang/opencoarrays/Makefile
Great, committed the rest, sorry for the mistake.