Hi,
in Sabayon we are providing our users kde-l10n split ebuilds (as well
as openoffice-l10n ones) since a long time ago. The split has many
advantages and follows this rationale:

- Gentoo seems to follow the rule to not split packages over what
upstream does, the inverse condition (merge more packages into one)
should be respected too.
- Gentoo is about choice, providing this alternative way can be a plus
and introduce several advantages for users:
   source world advantages:
   - it makes possible to add/remove kde-l10n packages without having
to recompile all the localizations every time
   - "backward compatibility" kept through kde-l10n-meta ebuild
   - ebuild maintenance requirements don't grow with ebuilds' number
growth (5 lines bash script does enough magic on version bumps)
   - scalable disk space requirements without the annoying recompilation phase
  binary world advantages:
   -  ability to provide split packages makes possible to support all
the languages without forcing users to install a 500+ mb package
   - reduced install time
   - reduced disk space
   - improved administration power and TCO (add/remove support for a
language becomes trivial)
   - scalable disk space requirements

Invalid complaints ( :P ):
- improved maintenance costs: it's not true, scripting exists for a reason
- There are USE flags: use flags tuning is not possible in binary
worlds and always forces users to recompile (which is not always
possible)

Reference ebuilds:
- http://svn.sabayonlinux.org/overlay/kde-base
Quick and dirty update script example:
- http://svn.sabayonlinux.org/overlay/kde-base/do-kde-l10n.sh
Example of kde-l10n-meta ebuild:
- 
http://svn.sabayonlinux.org/overlay/kde-base/kde-l10n-meta/kde-l10n-meta-4.2.2.ebuild

Opinions?

-- 
Fabio Erculiani
Sabayon Linux - http://www.sabayon.org
My life, my stories. Itsme - http://www.itsme.it

Reply via email to