The media-sound/musescore build fails if done in parallel, i.e. "make -j4" fails with random dependency issues but "make -j1" works. This is on an AMD64 X2 processor. Maybe this should be reported upstream, but for now the ebuild should filter out parallel makes. Thanks, Sebastian Reproducible: Always
Keeps working here with -j3 (dualcore machine)..
Fails on "-j4" but works on "-j3" here too. Hard to tell if that is always the case though. It probably should filter parallel builds as Sebastian suggested. This is on an Intel Quad Core.
+ 10 May 2009; Samuli Suominen <ssuominen@gentoo.org> + musescore-0.9.2.ebuild: + Remove built_with_use and force -j1 wrt #225457.
(In reply to comment #2) > Fails on "-j4" but works on "-j3" here too. Hard to tell if that is always the > case though. It probably should filter parallel builds as Sebastian suggested. > > This is on an Intel Quad Core. > Failed here too (I own a quadcore now)
Forcing -j1 is not a fix, it's a temporary workaround. Please leave such bugs open until you or upstream (you did report it right?) fixes the Makefile. http://blog.flameeyes.eu/2008/10/29/for-a-parallel-world-theory-lesson-n-2-handling-broken-ebuilds
No, haven't reported it upstream yet. Guys, please try =media-sound/musescore-0.9.4 so that you remove or comment the MAKEOPTS line from it, and report back.
(In reply to comment #6) > Guys, please try =media-sound/musescore-0.9.4 so that you remove or comment the > MAKEOPTS line from it, and report back. musescore-0.94 suffers from the same (tested on an old single-core amd64 with -j2 and -j1, only -j1 passed). In the mean time, musescore-0.95 is out, and while trying to update the 0.94 ebuild for it, I noticed that 0.95 started to fail as well with -jX (X > 1) when I was making a patch for it to apply a similar build-fix as there is for 0.94.
(In reply to comment #7) > > In the mean time, musescore-0.95 is out, and while trying to update the 0.94 > ebuild for it, I noticed that 0.95 started to fail as well with -jX (X > 1) > when I was making a patch for it to apply a similar build-fix as there is for > 0.94. > Could you post your musescore-0.95 ebuild? So we can at least install it with -j1. I edited the ebuild for ver 0.95. It compiles but file collisions occured with qtscriptgenerator. I guess your patch is for disabling qtscriptgenerator within ebuild. Could you post it, too?
(In reply to comment #8) > Could you post your musescore-0.95 ebuild? So we can at least install it with > -j1. I edited the ebuild for ver 0.95. It compiles but file collisions occured > with qtscriptgenerator. I guess your patch is for disabling qtscriptgenerator > within ebuild. Could you post it, too? I just removed the musescore-0.9.4-noscript.patch (since >=qt-4.5 and >=cmake-2.6 are stable and i don't see why awlplugin should not be built) from ebuild (copied musescore-0.9.4.ebuild -> musescore-0.9.5.ebuild) and added my own patch which removes qtscriptgenerator from build process. Builds musescore with -j1 successfully. Quick test also shows the program working. (Btw, SRC_URI has changed; previously mscore-${MY_P}.tar.bz2 and now it is MuseScore-${PV}.tar.bz2 ) Hope this helps. Pasting (un)conveniently the patch used for getting rid of the file collisions: --- mscore/CMakeLists.txt.old 2009-11-16 17:16:57.000000000 +0200 +++ mscore/CMakeLists.txt 2009-11-16 17:17:26.000000000 +0200 @@ -196,7 +196,7 @@ endif (NOT MINGW AND NOT APPLE) set (subdirs awl osdabzip rtf2html mscore share manual demos - scriptgen singleapp fluid) + singleapp fluid) if (MINGW OR APPLE) set (subdirs ${subdirs} portmidi)
Created attachment 210668 [details] unified diff for media-sound/musescore-0.9.5 This is the diff of media-sound/musescore-0.9.5 which is made against the one in portage media-sound/musescore-0.9.4. Also needs the musescore-0.9.5-build.patch. I hope to see this 0.9.5 ebuild in portage.
Created attachment 210669 [details, diff] musescore-0.9.5-build.patch This is the patch used for getting rid of the file collisions. Thanks to Jyrki for providing the patch on comment #9.
This isn't a problem in 1.1 now in CVS.
