getting rid of kde-l10n

[Ana Guerrero]

On Tue, Feb 02, 2010 at 04:00:57PM -0600, Rex Dieter wrote:
> On 02/02/2010 03:45 PM, Albert Astals Cid wrote:
> > A Dimarts, 2 de febrer de 2010, Rex Dieter va escriure:
> >> Hear me out. :)
> >>
> >> I'm not proposing anything yet, but would like to hear about the
> >> pros/cons of
> >> 1.  status quo: shipping translations separately to software
> >>
> >> 2.  shipping translations alonside (ie, in the same release tarballs)
> >> the software.
> >>
> >> How hard would it be to work toward something closer to 2 for KDE SC?
> >
> > What would be the benefit?
> 
> I know it's been this way for like forever.  Just wondering what kinds 
> of attachments there are, and pros/cons of each approach.
> 
> Honestly, part of this comes from my being a distro packager, but 
> keeping these separate is ... a bit painful(1).  Also, fedora's release 
> team keeps asking my why kde keeps translations separate, and I honestly 
> don't have a good answer for them.
> 
> Seems only the SC portions keep things separate, I mean extragear 
> releases bundle software/translations together, so it's not like it's 
> something new or different.
>

With my packager hat on, I only see benefits, and not only because
you can provide translations updates without having to update anything else.
This is useful specially when you distro is close to release and frozen.

First, most of the users are only interested in translations of their own
language(s) and the easiest is just install the package kde-l10-XX from their 
distro and get done with it. 

If you ship e.g. the kdeedu tarball with the translations, you have several
options:
- ship every app with their translations. This is not a big pain for
  packagers but it will make users have a lot of translations installed 
  they are not interested in.
- ship the translations in another package, per application (kdeedu-l10n) or 
  per module (marble-l10n). Again you make users install translations they 
  are not interested in and also have them installing several translation 
  packages. Packages have here aditional work having to care for way more
  translations packages than before.
- ship the translations in another package for every language:
  kdeedu-l10n-es, kdeedu-l10n-fr, etc. Again, more work for packagers and for
  users.
- I will omit the possibility of translation package for application and
  language (marble-l10n-fr), because I think it is plain crazy.

You could say that with the current approach, you have translations installed
that you do not need if you use only a couple of KDE apps, but it seems a
better approach installing 30 MB of translations for a couple of KDE apps that
make KDE users install 600 MB of translations.

The handling of extragears translation is not exaclty the way to follow here.
In every release packager have to update the languages, and users in some
cases do not know they have to install a translation for this package
manually. For them, it is just another kde app...

Said all this, could you tell me the benefits of shipping translations
alonside? I just do not see any.

Ana