[twg-tds] [gaulle@idris.fr: a TDS question]

Bernard Gaulle gaulle at idris.fr
Tue Jun 24 13:29:52 CEST 2003

Karl Berry writes:
 >     Why not have a language.cnf file allowing
 >          to set TEXINPUTS.language_name definition lines? 
 > That sounds like a very interesting idea to me. I wonder about a couple
 > of things, though:
 > 1) the mapping between \language numbers and names is not constant,
 >    right?  For instance, one system might have German as \language1, and
 >    another have French.  Does any information get dumped in the format
 >    which could help determine this?  

right, language numbers and names differ from one format to another.
\l@<language_name> (which give the numbers) are dumped.
At run time, \languagename, provides the _current_ 
language name, as far as i know the various multilingual packages.

 > 2) interaction between the existing paths and this new one.  I guess it
 >    would get read first.  

Well, in fact, i think we are talking about 
relative search paths. I explain:

Given the example of bibliography and calling plain.bst 
(which is usually at texmf/bibtex/bst/base/plain.bst).
In a multilingual environment one should want to access, e.g.
english/plain.bst or french/plain.bst or german/plain.bst ETC.
So we can imagine to have such files in ${TEXMF}/bibtex/bst/ if they
are not related to any format, otherwise it could be in

So, probably, the question is to be able to say: 
don't search base/plain.bst but <language>/plain.bst
OR better search first <language>/plain.bst.

In that scheme TEXINPUTS.language_name is looking to have no interest.
Better, it could be some variable such as 

Just few ideas for discussion...


PS: using as example the file plain.bst is not a so good idea since
    it is inputed once per document. In fact the multilingual docs
    switch from one language to another and then input may differ (as
    formerly discussed with linguistic keyboards).

More information about the twg-tds mailing list