[Dict-common-dev] pspell, pspell-ispell, aspell and dictionaries-common

[Agustín Martín Domingo]

Rafael Laboissiere wrote:
> * Agustín Martín Domingo <agmartin@aq.upm.es> [2002-06-06 18:44]:
> 
> 
>>I forgot to add one thing here, ispell.el currently allows only 
>>iso-8859-1, iso-8859-2, and koi8-r as values in the ispell.el emacs 
>>entries, otherwise will fail. So it is not possible to use the 
>>"Coding-System" field unless we patch ispell.el. This is another reason 
>>why I prefer the other system.
> 
> 
> I do not think that this is a problem.  I can easily modify the
> build_emacsen_support subroutine in DictionariesCommon.pm, such that no
> emacsen entry is created when the "Coding-System" field differs from one of
> the three above.  What do you think?
>  


No entry also does not work, you need to put exactly one of those 
entries, otherwise will fail too,

(debian-ispell-add-dictionary-entry
   '("american"
     "[A-Za-z]"
     "[^A-Za-z]"
     "[']"
     nil
     ("-B")
     nil
     nil)
   "american")

will give an error

   set-variable: Value `(("american" "[A-Za-z]"
   ...
   does not match type repeat of ispell-dictionary-alist

as well as putting iso8859-1. This latter may also be a problem for the 
pspell entries, which use iso8859-1 instead of iso-8859-1.

Domenico, how critical is for pspell-ispell naming the charset as 
iso8859-1 or iso-8859-1?. Do we have a list of allowed charset names for 
pspell-ispell to try some mapping? Otherwise we can always use something 
like a 'Pspell-ispell:' two elements field in the info file:

Pspell-ispell:sv iso8859-1

without any sort of mapping.

 From the other message:
> 
> Okay, I buy your arguments and also think that changing installdeb-ispell is
> the way to go.  There will be no problems with multiple hashes, since the
> Perl code in installdeb-ispell can parse the multiple definitions in the
> ispell-info file.  I will try to take a look at the implementation.

I was also looking at the installdeb-in source and seems not difficult 
to make it behave as we want, as a matter of fact it already parses the 
info file for testing its presence.

One more thing we should also consider when doing that, two different 
info-ispell definitions can correspond to a unique hash (e.g. castellano 
and castellano8) giving different emacs entries, so one possibility os 
to allow some definitions not having a pspell field. Although this is 
probably not a problem if there are pspell fields in the two entries, 
since the later pspell-ispell file will overwrite the previous one with 
the same info.

Cheers

Agustin,

-- 
=====================================================================
Agustin Martin Domingo, Dpto. de Fisica, ETS Arquitectura Madrid,
(U. Politecnica de Madrid)  tel: +34 91-336-6536, Fax: +34 91-336-6554,
email:agmartin@aq.upm.es, http://corbu.aq.upm.es/~agmartin/welcome.html