Currently the call to eqmake4 in src_configure() is totally useless because the top-level Makefile will later re-run qmake, thus destroying eqmake4's effects. Upstream's Makefile sucks a lot, so the solution is to rewrite most of its default rule in a way that it recursively invokes itself on the subdir 'src' and do nothing else. While at it, I changed the generation of svn_revision.h to actually report the correct SVN rev when the ebuild is a snapshot (the revision number is extracted from ${PV}). Furthermore the code should now be able to handle both regular releases (i.e. -x.y.z without _pXXXX components) and unofficial svn snapshots (e.g. -0.6.7_p2872), without having to change ${S}, ${SRC_URI}, etc... Finally, I fixed a small bug. Look at the following snippet: [[ -f *.qm ]] || lrelease ${PN}_en_US.ts Filename expansion is not performed inside [[ ]] -> *.qm remains as it is -> bash tries to literally match it -> the test is always false because there's no file named "*.qm" -> the en_US translation is always generated.
Created attachment 187987 [details, diff] diff against 0.6.7_p2872-r1
CC'ing yngwin since he is smplayer's maintainer.
No need to CC me, as I'm already in media-video herd. ;-) Thanks for the patch, it looks great!
(In reply to comment #3) > No need to CC me, as I'm already in media-video herd. ;-) > Oops! Sorry about that... :)
ping? the usage of 'make' instead of $(MAKE) in the makefiles breaks the build on fbsd and this patch seems to fix it.
Applied in smplayer-0.6.7_p2872-r2.ebuild, thanks for your contribution!
(In reply to comment #6) > Applied in smplayer-0.6.7_p2872-r2.ebuild, thanks for your contribution! Are you going to upstream these patches?
(In reply to comment #7) > Are you going to upstream these patches? > Yes, I intend to.