Re: Packaging very large i18n material
> 2). I've build the koffice-i18n packages in the same way that the kde-i18n
> packages are. That avoids 37 source packages at least but only gives you the
> languages.
Sure, it was always going to come from one source package. It's the 37
*binary* packages I was worried about, which is what the end users have
to deal with.
> Also it scales much better because you just have to take the
> kde-i18n package's debian dir and change it accordingly to spit out
> koffice-i18n-* packages.
This is certainly true, but then again being easier on packagers doesn't
necessarily mean it's the best solution for debian.
> That's the way things are handled since koffice 1.2
> and if we continue this way we won't break upgrading for those who have
> koffice installed on woody.
While I'm not arguing against option (2) specifically, I disagree with
this reasoning. Debian should not in general abide by whatever decisions
are made by third-party packagers for unofficial repositories, since
often such third-party packagers have different priorities, e.g.,
"let's make this software available right now" as opposed to "let's make
sure this software integrates well with the rest of debian". I'd rather
make sure we make the best decision when uploading to sid than just
continuing the status quo defined by unofficial packagers.
Again I'm not arguing against you, your packaging or option (2); I'm
arguing against your logic. :) It's looking at the moment that the 37
binary packages is indeed close to the optimal decision in these
circumstances.
Ben.
Reply to: