I agree we need to support this, except doing so presents a very realistic and major issue. The way strings are handled in glade/gtkbuilder is very different than the way strings are handled in bonoboui. However, bonoboui has already been using the .ui extension for its ui xml description files much longer, and there are still a significant number of applications using them. The changes in the linked branch would break those applications at the same time, so I don't think we can land it yet. And I'm not sure what the best way to deal with this unfortunate choice of file extension is at the moment. I would rather avoid parsing the XML to figure it out, as well. If they all had proper mime types defined, we could probably check that instead of just using a simple regex against the file name extenions, as we do now. I think there's a useful implementation of xdgmime in PERL available, so that would at least help us avoid the performance hit for calling file --mime-type (or some equiv) for every possible file as well.
I suppose the 'mime type' way would be the most clean way to handle this, if it's not too heavy from a performance point of view. I've even seen some GTKBuilder files keeping a .xml extension, so the extension is not an option IMHO. Anyway, the current option to explicitely prefix the files in POTFILES.in is not so bad, even if a little verbose...
I agree we need to support this, except doing so presents a very realistic and major issue. The way strings are handled in glade/gtkbuilder is very different than the way strings are handled in bonoboui. However, bonoboui has already been using the .ui extension for its ui xml description files much longer, and there are still a significant number of applications using them. The changes in the linked branch would break those applications at the same time, so I don't think we can land it yet. And I'm not sure what the best way to deal with this unfortunate choice of file extension is at the moment. I would rather avoid parsing the XML to figure it out, as well. If they all had proper mime types defined, we could probably check that instead of just using a simple regex against the file name extenions, as we do now. I think there's a useful implementation of xdgmime in PERL available, so that would at least help us avoid the performance hit for calling file --mime-type (or some equiv) for every possible file as well.