Tried to compile dev-libs/libowfat-0.28-r1 and got the following error: /usr/bin/diet -Os gcc -c socket/socket_remote4.c -I. -march=native -O2 -pipe -fomit-frame-pointer socket/socket_remote4.c:11:23: error: havescope.h: No such file or directory It appears that make isn't creating havescope.h before trying to compile socket_remote4.c which relies on it. I am assuming that it may work with some versions of make(GNU Make 3.82) or something like that. A patch is attached to fix the makefile as well as a patch to the ebuild. The ebuild patch also disables dietlibc as I assume most gentoo users would rather use their standard libc instead of adding a new library. I would make it a use flag as in bug 97185 but there isn't one for dietlibc and I am not sure the best way to go about doing that. Reproducible: Always Steps to Reproduce: 1. emerge dev-libs/libowfat-0.28-r1 2. 3.
Created attachment 286523 [details, diff] Ebuild patch
Created attachment 286525 [details, diff] libowfat-0.28-GNUmakefile.patch
Anything I can do to get this submitted faster? Currently it won't compile for I am guessing most gentoo users.
Can't reproduce it ... even with MAKEOPTS="-j" and trying a few times
It isn't an issue with -j. It fails for me even with no -j and when trying to build it on its own not via portage. The makefile does not state that havescope.h is a dependency of socket_remote4.c even though it should. havescope.h is a file generated by the makefile. The real question is why it doesn't fail on some systems without the patch. I am guessing that it has something to do with the makefile using lines like: SOCKET_OBJS=$(patsubst socket/%.c,%.o,$(wildcard socket/*.c)) to generate the list of files to compile. On your system it is most likely compiling something else that requires havescope.h before socket_remote4.c while on mine it is not. Either way not specifying the dependency seems wrong and seems to be a problem on some systems. I am attaching a new version of the patch that is done properly unlike my first one.
Created attachment 290589 [details, diff] libowfat-0.28-GNUmakefile.patch
+ 22 Mar 2012; Patrick Lauer <patrick@gentoo.org> + +files/libowfat-0.28-GNUmakefile.patch, libowfat-0.28-r1.ebuild: + Compile fix for #409139 #383031 So I was able to reproduce it - diet useflag seems to be one of the triggers, very strange. Seems to build fine now, thanks for the patch!
