specspo and PackageKit
skvidal at fedoraproject.org
Wed Sep 24 14:23:30 UTC 2008
On Wed, 2008-09-24 at 10:20 -0400, Bill Nottingham wrote:
> seth vidal (skvidal at fedoraproject.org) said:
> > On Tue, 2008-09-23 at 17:20 -0400, Bill Nottingham wrote:
> > > I'm just trying to comprehend how we'd generate this information for
> > > the repodata in any sort of sane manner, and failing miserably. You'd
> > > need to have every run of createrepo check out some source controlled
> > > translation directory, and munge it with intltool (aiyeeeeeeeee) into
> > > a separate translated XML file.
> > No, you don't need to do it on every run of createrepo. You can do it
> > after the fact, if you want and then hit the repo with modifyrepo to add
> > the translation data. You could even have a separate
> > repomd-translation.xml which lists out the translation files available
> > and their checksums, etc.
> > The whole point is specspo isn't useful on its own anymore. I understand
> > things we may do from here will take some work but specspo just takes up
> > space w/o providing any functionality at all afaict.
> You're specifically architecting a repodata-based solution to replace
> just what it does now. How is that 'w/o providing any functionality at
No, I was offering a way of doing it so you only got the translations
you needed for the language you were using. Additionally, I was trying
to focus on a solution which allowed you to have translations for both
the installed and available packages.
In my ideal world things would work like this:
1. translations live in the repodata in separate files per language
2. yum downloads the translations for the lang you're running in
3. when you install a package it takes the translations from the
repodata translation and stuffs them in the rpmdb alongside the stock
info for the pkg.
1 and 2 should be doable.
3 is tricky w/o support from rpm, but not impossible.
More information about the fedora-devel-list