tags 595260 pending thanks On Thu, 2010-09-02 at 11:20 -0400, Aaron M. Ucko wrote: > Package: salome > Version: 5.1.3-10 > Severity: serious > Justification: fails to build from source > > The automatic build of salome on i386, which appears to have been in > parallel, ran into what looks like a race condition between cleaning > subtrees after building their contents and building and installing > documentation:
Thanks Aaron, I noticed this. What's the best way around it, while preserving the performance of a parallel build? In the current version on alioth, build-indep-stamp is independent from build-arch-stamp, so I can't make one depend on the other to force -arch-stamp to complete before -indep-stamp starts. Is there a way to put a "makefile barrier" so all of the threads catch up before moving on? I'm not finding what I'm looking for in the make manual http://www.gnu.org/software/make/manual/html_node/Parallel.html . Ah, just hit on a solution: I'll have the documentation build in a separate directory. I'm still interested in the answer to the above though ("makefile barriers"). Just implemented and pushed this to alioth, hence "pending" above. > Here are some relevant excerpts from the log > (https://buildd.debian.org/fetch.cgi?pkg=salome;ver=5.1.3-10;arch=i386;stamp=1283419999 > ), > in order: [snip] > I see failures on several other architectures as well, quite possibly > due to the same issue (though I haven't actually checked); at any > rate, could you please address it? Just checked: Sparc, IA64 and Alpha all attempted with -j 2 and failed; PPC and S390 used -j 1 and succeeded. Hmm, Sparc and IA64 fail in a different place, which doesn't appear to be related to the jobs issue. I'll file a separate bug. Thanks again, Adam -- GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Engineering consulting with open source tools http://www.opennovation.com/
signature.asc
Description: This is a digitally signed message part