hi,
I seems that
make -j16 all perl-build
fail in a lot of different place.
Could you try to investigate. Maybe it is a bug to report to automake
Bastien
Parallel make unsafe
Re: Parallel make unsafe
We can reproduce the problem on a CentOS 5.9 system with make 1:3.81-3. On our Fedora system with make 1:3.82-13 it consistently worked fine. Perhaps its a bug in make that has already been fixed in recent versions of make.
-
- Posts: 467
- Joined: 2008-12-21T11:51:10-07:00
Re: Parallel make unsafe
I occurs about 1 on 10 try with make -j32 on a quad core.
I suppose it is a bug (or a feature) of automake.
It seems that coders does not depend (wait) on libmagickcore fully builded.
Bastien
I suppose it is a bug (or a feature) of automake.
It seems that coders does not depend (wait) on libmagickcore fully builded.
Bastien
-
- Posts: 467
- Joined: 2008-12-21T11:51:10-07:00
Re: Parallel make unsafe
Any news of this ?
Re: Parallel make unsafe
The parallel make bug has existed for many years and it appears likely to be a race condition triggered by make or automake. If you think its an ImageMagick bug, let us know and we will investigate further.
-
- Posts: 467
- Joined: 2008-12-21T11:51:10-07:00
Re: Parallel make unsafe
Please see automake bug here. same code is in coder.
We do not need to workarround it (workarround is ugly and probability of faillure is low). But I think we should document the problem
Upstream bug
http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7328
Could you please also add comment to the upstream bug report and track it ?
Bastien
We do not need to workarround it (workarround is ugly and probability of faillure is low). But I think we should document the problem
Upstream bug
http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7328
Could you please also add comment to the upstream bug report and track it ?
Bastien