[tex-live] German exception wordlist (was Re: Re: hyph_size)

[Petr Sojka]

On Fri, Aug 05, 2005 at 10:06:33AM +0200, tex-live-request at tug.org wrote:
There is a discussion about adding dehyphtex.tex to texlive on texlive
list today:
 
> Today's Topics:
> 
>    2. Re: hyph_size (Thomas Esser)
>    3. Re: hyph_size (Frank K?ster)
>    4. Re: hyph_size (Frank K?ster)
>    5. Re: Re: hyph_size (Karl Berry)
>    6. Re: Re: hyph_size (Frank K?ster)
>    8. Re: Re: hyph_size (Karl Berry)
>   10. Re: Re: hyph_size (Werner LEMBERG)
...
> > Are changing line breaks a concern for German?  Or is this something
> > simply loaded by users?
> 
...
>     It is loaded by users who want it, after \usepackage{german}.  
> 
> Ok, good.  I added it to TL (package hyphen-german).  Thanks.
> 
>      BTW, why after - and what about babel, which activates lots of
>      stuff only at \begin{document}?
> 
> Because the file simply contains a big \hyphenation {} pattern,
> nothing else -- it doesn't explicitly switch to German.
> 
> > - and what about babel, which activates lots of stuff only at
> > \begin{document}?
> 
> You have to load the file where German is active, perhaps with
> 
>   \begin{otherlanguage}{german}
>     \input{dehyphtex}
>   \end{otherlanguage}

Comment/suggestions:
-- as dehyphtex.tex is ISO-Latin1 encoded file this may end
   in portability problem when using different input encoding.
-- as 
   -- the list is rather huge (which actually shows that it is
   time to regenerate the patterns from scratch similarly as
   for English), 
   and 
   -- is consulted by TeX for every word to be hyphenated
   and
   -- may cause troubles with the size of hyph_size
   I'd suggest as a temporary solution either 
   a) making new version of dehypht.tex such that dehyphtex.tex is 
   converted to additional patterns in such a way that
   -- . is addres to the both beginning and end of every word exception
   -- 8 is added at the wrong positions
   -- 7 is added at the good positions.
   or
   b) using patgen with dehypht.tex patterns and dehyphtex list as
   input and generationg two new passes covering the new words
   (slightly more tricky but more space effective solution than a)).

This would have advantages:
-- it is much quicker (no need to read dehyphtex.tex in every run
   of German document), leaving exception dictionary empty for user's
   exceptions
-- it is more portable (using the encoding of dehypht.tex) and encoding
   problems free
-- people would not have to think of adding dehyphtex.tex into 
   every German document
WDYT?
--ps