Re: language specifics

[Karsten Tinnefeld <tinne@goedel.cs.uni-dortmund.de>]

> The main problem is in the weird ones. I have no problem with the "e
> and alike ones, but using them for compound words and quotes is a bad
> idea, since context has dedicated alternatives for those, which also
> have options. To what extend do we need to satisfy habits and features
> from for instance latex?

It is a sensible argument that many of the abbreviations I've been 
discussing to you in the last few days may be kind of complicated crap, 
let alone of no use to a lot of languages.

However, the functionality provided appears to me to be quite important 
- at least as long as we still build on 7- or 8-bit input code sets. 
The german package for that instance has evolved for the last twelve 
years, so that I expect the basics to having been settled. On the other 
hand, all but the very simplest ligatures are defined more or less 
arbitrarily, and explaining plain paragraph source to a wysiayg user is 
a nontrivial task.

Thus, if context has even more powerful tools that can cope in all 
necessary cases, maybe this is the time to consider turning to a more 
reasonable source encoding. In this case, my questions are:

a) Do we still have to account for 7-bit ascii?
b) How do we encode the various ways of hyphening and dash-ing in a most 
   non-ideosyncratic way?

Karsten
-- 
Karsten Tinnefeld                       tinnefeld@ls2.cs.uni-dortmund.de
Fachbereich Informatik, Lehrstuhl 2                   T +49 231 755-4737
Universität Dortmund, D-44221 Dortmund, Deutschland   F +49 231 755-2047