Peter Dyballa <Peter_Dyballa@Web.DE> writes:

> Am 30.08.2006 um 22:53 schrieb Michaël Cadilhac:
>
>> Peter Dyballa <Peter_Dyballa@Web.DE> writes:
>>
>>> Am 30.08.2006 um 21:44 schrieb Sébastien Vauban:
>>>
>>>>> (defun ispell-get-otherchars ()
>>>>>   (replace-regexp-in-string "-" "" (ispell-get-decoded-string 3)))
>>>>
>>>> I did add your function, but... no, it does not work... I still
>>>> got the coumpound words underlined...
>>
>> Strange... I use CVS Emacs, what version are you running?
>
> 21.3.50, 22.0.50, 23.0.0.
>
>>
>> Where did you put it?  If it's in your .emacs, did you put a (require
>> 'ispell) before?
>
> No. But I can tell you that your lines change the value of OTHERCHARS,
> from default "[-'^`\".@]" (in a version running some  time) to "[-']".

Gosh! How come? In ispell.el, (ispell-get-otherchars) is

(defun ispell-get-otherchars ()
  (ispell-get-decoded-string 3))

What I just did is to REMOVE the "-" from this string. I really don't
see how (replace-regexp-in-string) could lead to the result you say it
leads!


However, I've made a big mistake so far, you guys were talking about
ispell, will I had flyspell in mind. Flyspell using ispell-get-word
(or so), the underlining is rightly done with the fix I propose. But
M-x ispell-buffer using ispell-get-line (or so) and calling ispell on
it, the result is not the same.

> I'm not sure whether it's an Emacs thing to decide where a word ends.

Depends on the case. The present one is not determined by Emacs (my
mistake).

> IMO Emacs passes a region or a buffer to the ispell process which
> then reads through this. And so it's ispell then, that decides where
> a word ends. Since I think ispell's not correctly set up ...

You're completely right, my bad.

> In Fink Martin Costabel is the maintainer of this package; in
> francais.aff this is recorded: Copyright 1999, Christophe Pythoud et
> GUTenberg.

From here, I consider I don't have the necessary knowledge to help,
sorry.

> Would it work to subtract - from OTHERCHARS in
> ispell-dictionary-alist-3?

No, that's what my proposition did, even if it seems not to work with
you (any idea why?).

However, this solution is maybe equivalent (but at another level,
ispell's one) to remove ``boundarychars [-]'' from francais.aff.

-- 
 |      Michaël `Micha' Cadilhac   |  Isn't vi that text editor with        |
 |         Epita/LRDE Promo 2007   |   two modes... One that beeps and      |
 | http://www.lrde.org/~cadilh_m   |     one that corrupts your file?       |
 `--  -   JID: micha@amessage.be --'           -- Dan Jacobson         -  --'