Subject pretty much says it all... Patch attached, error message below: USE="doc" ebuild gnome-panel-2.5.93.ebuild unpack >>> md5 src_uri ;-) gnome-panel-2.5.93.tar.bz2 >>> Unpacking source... >>> Unpacking gnome-panel-2.5.93.tar.bz2 to /var/tmp/portage/gnome-panel-2.5.93/work configure.in:26: warning: AC_ARG_PROGRAM invoked multiple times gtk-doc.make:7: GTK_DOC_USE_LIBTOOL does not appear in AM_CONDITIONAL gtk-doc.make:44: ENABLE_GTK_DOC does not appear in AM_CONDITIONAL gtk-doc.make:135: ENABLE_GTK_DOC does not appear in AM_CONDITIONAL !!! ERROR: gnome-base/gnome-panel-2.5.93 failed. !!! Function src_unpack, Line 51, Exitcode 1 !!! (no error message) Reproducible: Always Steps to Reproduce:
Created attachment 27743 [details, diff] ebuild patch This patch just adds a test and optional sed line to the ebuild...
That would work for now, but the real solution would be to fix the makefile script to not load the gconf defaults. Maybe you could patch that up ? blocker ? -> easy workaround is present
Why not just do the gconf sed on the Makefile after configure is done? Saves time (no aclocal/automake needed) on the compile, and doesn't require this hack either... I set to blocker since it keeps anyone from emerging that doesn't have doc in their USE flags, or gtk-doc emerged. Counts as blockage in my book....
true, the hack will suffice for now. But what i'd like to see is a full fix... oh and why we do the sed before is that in general we consider it a bad habit to sed in generated files (patching should happen in _unpack, configure in _compile)
I understand the reasoning behind sed/patches in src_unpack, but it just seems that in this case it would make more sense in src_compile... I can look into a full fix, but no promises... Got a lot on my plate atm..
ok for the time being i moved the sed to src_configure in 2.6.0, but i'd like to see a real fix here someday. Please test
*** Bug 45451 has been marked as a duplicate of this bug. ***
i'm going to assume it works, still waiting for the clean patch though.