svn commit: r486753 - head/cad/gmsh

[Joseph Mingrone]

Alexey Dokuchaev <danfe at freebsd.org> writes:

> On Thu, Dec 06, 2018 at 11:55:00AM +0000, Joseph Mingrone wrote:
>> New Revision: 486753
>> URL: https://svnweb.freebsd.org/changeset/ports/486753

>> Log:
>>   cad/gmsh: Update to 4.0.6

>>   Upstream change log:
>>   https://gitlab.onelab.info/gmsh/gmsh/blob/master/CHANGELOG.txt

>> -BROKEN_FreeBSD_12_i386=	fails to build on FreeBSD 12 i386 due to a clang bug: https://llvm.org/bugs/show_bug.cgi?id=21903
>> -BROKEN_FreeBSD_13_i386=	fails to build on FreeBSD 13 i386 due to a clang bug: https://llvm.org/bugs/show_bug.cgi?id=21903

> It would be nice to explicitly acknowledge in the commit log that it builds
> fine on i386 now, and these lines were not removed by accident.

>> -MAKE_JOBS_UNSAFE=	yes
>> -

> I didn't find anything about parallel build fixes in the changelog, and
> commit message does not mention it as well.  It is expected that when -jX
> unsafe issue it fixed (and thus MAKE_JOBS_UNSAFE removed), an approximate
> explanation is provided in the commit log.  People often try parallel
> builds locally and they seem fine, while the problem might just not so
> easily manifest itself.

> ./danfe

I agree in both cases that the commit log fell short.  This was careless
on my part, so pointy hat to me.

According to the submitter (see the referenced PR, 233618), he tested
this in poudriere and the build now succeeds on 12/13 i386.  As for the
removal of MAKE_JOBS_UNSAFE=yes, again, this should have been documented
by both the submitter and me.  However, given it was added over 5 years
ago and PR 182125 says 'How-To-Repeat: Build on a MP machine', I am
inclined to leave things as they are for now.

--
Joseph
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 962 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/svn-ports-all/attachments/20181206/3492defc/attachment.sig>