On Fri, Oct 08, 2010 at 11:32:04AM +0200, Jérémy Bobbio wrote: > On Fri, Oct 08, 2010 at 10:23:19AM +0200, Mike Hommey wrote: > > So, I was able to reproduce the issue, and the blame is shared between both > > iceweasel and torbutton. The problem here is that torbutton overrides > > one of the core components of iceweasel (exthandler), and as such > > iceweasel uses torbutton's. After torbutton removal, iceweasel fails to > > initialize the exthandler component because it disappeared, and thus > > can't start. > > > > I think this is a core problem in iceweasel/gecko (which also applies to > > other gecko-based applications), and I think the best we can do in a > > timely manner, without waiting for an absolute right fix is to use > > dpkg triggers. > > > > The way it would work is that dangerous extensions such as torbutton > > would send a trigger to iceweasel and other packages it would break > > (icedove, iceape...), which name we'd agree upon (how about > > xulrunner-autoreg?) and iceweasel and other packages would do the right > > thing. > > I might not be knowledgeable enough to have a clear idea of what such > trigger will do. Would you be kind enough to give more details on what > you think? > > Are you aware of any other extensions that would benefit from such > trigger? > > I just had a quick look at every packages containing an install.rdf > file, and I have not spotted anything looking like core components > overidden by other extensions. I have also done quick tests for > foxyproxy, iceweasel-downthemall, iceweasel-vimperator, > mozilla-imagezoom, mozilla-livehttpheaders, xul-ext-greasemonkey and > xul-ext-webdeveloper. None exhibits the same behaviour as torbutton.
Maybe for only one extension using triggers would be bloated. Let's already wait a bit if I can come up with a patch fixing the real issue. If that patch is small enough we can probably get it in for the release. Otherwise, the postrm or prerm script in torbutton could touch /usr/lib/iceweasel/autoreg and /usr/lib/iceape/autoreg if they exist (and only if they do) Cheers, Mike -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
