language.dat: Should we just enable all the patterns?

[frank@kuesterei.ch (Frank Küster)]

Hi all,

I didn't want to raise this question for tetex-2.0.2, but since we have
that bug #235401, I decided to do it:

What about kicking out all my nice code for handling language.dat, and
all debconf related language.dat stuff, and simply enabling _all_
patterns we ship?

I never took this possibility into account before, just because it's
always been like that: It was necessary to enable the patterns one
wanted. But in fact there is little reason for that. With only the
required patterns (american and nohyphenation) and all debconf-managed
patterns deselected, I get

$ find /var/lib/texmf/web2c/ -type f -name "*fmt" | xargs du -hc  | grep total
8.4M    total

With all debconf-managed patterns enabled, except for norwegian where I
chose nohyphb, and ukranian where I took the first possibility, I get

$ find /var/lib/texmf/web2c/ -type f -name "*fmt" | xargs du -hc  | grep total
13M     total

This is a difference of 4.6 Mbyte. That's not nothing, but with today's
harddisk sizes I think it isn't really important. The only thing we have
to decide (and document in README.Debian) is which of the options to
chose for languages with multiple patterns, but equal babel names,
currently norsk and ukranian.

I guess it will save us bug reports, and work to be done (actually I
found on error in my code this morning, though it doesn't seem to be the
reason for the bug).

Comments?

Regards, Frank
-- 
Frank Küster, Biozentrum der Univ. Basel
Abt. Biophysikalische Chemie