Created attachment 514778 [details, diff] Patch: fixes RDEPEND (a) RDEPEND Contrary to the logic of the RDEPEND definition, the USE=gtk does NOT imply pygobject being present, and therefore it should be added unconditionally as a run-time dependency. Consequently, the very same dependency can be removed from the ncurses? guarded list. Patch attached. (b) /usr/portage/net-misc/wicd/files/wicd-init-sve-start.patch This patch in FILESDIR replaces 'start' in the initscript by 'sve_start', but 'sve_start' is nowhere to be found on my system, even after running 'emerge wicd'... What is this 'sve_start'? Where should it come from?
Created attachment 514780 [details, diff] Patch: fixes RDEPEND (correct syntax) Oops... Messed up the syntax. Sorry!
On a closer look, it seems there's more than just pygobject missing from RDEPEND... Trying to run wicd-gtk chokes on 'import gtk', which tells me that pygtk should probably also be included as a run-time dependency...
Can I not edit my own comments? How annoying... (-; In any case, the above should probably only apply with USE=gtk
AND, by "the above" I meant ONLY the single comment before, not *all* the previous comments. To be clear: the wicd daemon needs pygobject no matter what. Additionally, the gtk client (wicd-gtk) needs pygtk as well, but that depends on the USE flag.
The ebuild seems to already include pygtk, but only when USE=X is active. Assumptions violated: I'm building my system with USE="-X gtk"... BTW, I don't see any other effect of USE=X in the entire ebuild, *except* for selecting run-time dependencies between gtk and kde. This isn't exactly obvious or -for that matter- sane use of that flag...
Alright.. so, it seems that building with USE="X gtk" pulls in everything that's needed, and also partly explains the thinking behind the !gtk? guard on pygobject. I suppose the correct title for this bug would be something along the lines of confusing (mis)use of USE=X ...
https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=76cd886dac168af05cacd5f6d6aae790583fce27