Locale-specific sorting

[Michael Piotrowski]

Hi everybody,

Locale-specific sorting, as introduced in Lout 3.11, is certainly a Good Thing, 
especially compared with the ignorance of most other systems with respect to 
locale handling.

However, after being short of finally giving up on Lout because my references 
never got resolved, I think some more work has to be done in this area (I'm 
also 
willing to help here, at least with some "clever" comments :-)

The problem I experienced was that lout.li got not sorted the way that Lout 
expected it to be sorted, because my locale defines all non-alphanumeric 
characters to be sorted after the alphanumeric characters.

If LANG=C, I get the following result:

1&284.test.1    0       00001   0       1       test
17&681.test.1   0       00019   2121    171     test
 
Whereas if LANG=de_DE.iso88591, I get the following result:

17&681.test.1   0       00019   2121    171     test
1&284.test.1    0       00001   0       1       test

Lout obviously expects the first result, and consequently failed to resolve the 
references.

Solutions I could think of would be to always sort ``system files'' like 
lout.li 
in the C locale, so that the ordering is predictable, or to replace the 
text-based index with some kind of DBM file (which I'd prefer, but which is 
probably against Lout's design philosophy).

I'm posting this here to get some feedback and discussion, something which I 
consider very important for complex things like internationalization issues, 
before one implements another flawed solution.

Greetings

Michael


P.S.:  Besides criticizing, I should probably mention that I consider Lout a 
really great system.  After years of LaTeX hacking (i.e., not only writing of 
documents, but also of packages and classes), I'm really fed up and beginning 
to 
dispair.  Lout rescued me!

-- 
Michael Piotrowski                         <address@hidden>
Department of Computational Linguistics --- University of Erlangen, Germany
You know, basically, I'm no good at either linguistics or computer science.
                                                               --Larry Wall