[WOODY/WOODY+1] Excess Internationalizion Data
A rising problem I see in Debian is the excess i18n data that
every system has and the bad partitioning of it.
For the excess i18n data, I will point out that du /usr/share/locale
is 32M, 32M that I could do without. Even for the someone with
LANGUAGES="de:fr:en" (i.e. uses the German messages, then the French
messages, and English only if there isn't the first two) (using
the two languages that use the most of /usr/share/locale), he still has
28M of wasted space.
One solution (1) that has been used in Debian is storing those files in a
foo-i18n package, like tcsh-i18n. The problem is that it encourages
the proliferation of packages, and only solves the problem for English
speakers. (IMHO, as tcsh-i18n is < 300K of stuff, it should just be
merged back into tcsh.)
Another solution (2) is the creation of large packages named something
like lang-de which contains all the language specific files for one
language. This has several problems with size waste and difficulty
to keep current.
Sort of the cross between these two idea (3) has been used, i.e.
foo-lang type packages. This definetly encourages package prolification
and is very bad if done for every 2k message file.
My solution (ah, here's the complex technical solution that will require
reworking the deb format! (-:) is letting the sysadmin specify which
languages she will install when she installs the system, and then each
deb mark the language specific files and install only those. Possibly
with a special mark for documentation packages that are useless if you don't
use the language within. Problems here are the extra complexity and
the growth of the packages currently using schemes (2) and (3), like
fortunes and dictionary packages. It would also be nice to be able to
install files from a specific package for a unusual language.
So any opinions? Is it a solution in search of a problem? Would it
hinder or help further i18n'ing of Debian? (It's rather annoying to
me to have 18 Linux distributions, each for its own language, and I
really like to see Debian and other mainstream distributions start
removing the need for a special language distribution.)
--
David Starner - dstarner98@aasaa.ofe.org
Only a nerd would worry about wrong parentheses with
square brackets. But that's what mathematicians are.
-- Dr. Burchard, math professor at OSU
Reply to: