On Mon, 2011-05-02 at 11:52 +0200, Maarten Bosmans wrote: > Recently, I encountered some problems when enabling orc in some less > usual situations. > > When compiling with --enable-orc from a tarball generated from a > --disable-orc configured tree, the following error occurs. > make[2]: *** No rule to make target `pulsecore/svolume-orc-gen.c', > needed by `all'. Stop > I haven't really looked at a solution. Perhaps the nodist_ prefix for > some files inside if HAVE_ORC in src/Makefile.am should be dropped? > May be Colin needs to do his make distchecks with this situation, in > order to catch it earlier. There should not be a dist'ed tarball without those generated files. Whether they are used or not is a configure-time option, then. > Secondly, there is a problem when cross-compiling. The pkg-config > check for ORC is used to find the usual include files and linking > flags, but also to find the location of orcc. This is a problem, > because when configure is run with the correct configuration, such > that pkg-config finds the host package, it also finds the host orcc > (in the case of my mingw32 test, it finds orcc.exe), which is of > course useless in the build environment. Why is it useless in your environment? The files generated by orcc are architecture-neutral. Cheers, Arun
