On 10/11/2011 09:32 PM, Thomas Spura wrote: > The main *BIG* difference is, that draft symlinks the extension > *directory* and the script expects a install.rdf file below that. > This means, the symlinking needs to happen one step below that, so that > all files inside of the extension_id folder are symlinked, but the > install.rdf still needs to be a real copy, so it can be opened at build > time. > > Rationale: A directory symlink can't be resolved at build time, so we > cannot follow that symlink on build time. > > When that's changed, the scripts are working fine along each other. Could just use relative symlinks for the directories, so that they can be resolved at build time. > About "no dependency using" from above: > The dependencies will be added automatic with the scripts, so > to avoid pulling in e.g. seamonkey, when you only want to have > the firefox extension, there need to be one package for each > extension, which owns > /usr/share/mozilla/extensions/$browser_id/$extension_id. > > This way e.g. firefox-$extension would automaticalls require the > correct firefox versions but no seahorse, because that has to be owned > by seahorse-$extension - just as an example, but it would make sense. Yeah, I think there are two alternatives: a) one big package and no requires on specific browsers, b) split packages and each package requires a specific browser My draft used (a), but either way would work. > Kalev, does this make sense? Can this be integrated into the drafts? > I'll try to add those macros proposed there > into /usr/lib/rpm/macros.mozilla and see if they really work out. Could you just fork my draft and amend it? I am not sure I am sufficiently interested in properly finishing it up. -- Kalev -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
