On Tue, Nov 08, 2005 at 02:03:34AM +0100, Norbert Preining wrote:
> Hi Agustin!
>
> On Die, 08 Nov 2005, Agustin Martin wrote:
> > > My .emacs now contains:
> > > (custom-set-variables
> > > '(ispell-local-dictionary-alist (quote (("american8" "[A-ZÄÖÜäöüß]"
> > > "[^A-ZÄÖÜäöüß]" "[']" nil
> > > ("-B" "-d" "american" "-w" "öÖäÄüÜß") "~latin1" iso-8859-1)))))
> > >
> > > which works perfectly also with files containing "Gödel".
> >
> > Fine, thanks for the suggestion. And even better, octal codes seem to work
> > too (avoiding things be accidentally saved as utf-8) and to make things even
>
> Idea: Would it be possible to use the regexp teaching emacs additional
> word boundary chars (in my case the "[A-ZÄÖÜäöüß]") to automatically
> create the -w option?
>
> I guess, if you tell emacs that these chars are word components, ispell
> should also (automatically) know about it.
>
> The problem of course is that the one thing is an regexp, while the
> other is just a char list. But for many cases this could be done.
The problem, as you point out, is that can contain regexps that are not the
trivial A-Za-z that can easily be stripped (by the way, is a-z not missed in
your example?).
Another possibility I am thinking about is to do that in a per-encoding
basis, that is, with a list containing elements like
("iso-8859-1" . "àáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ")
so if ispell is used with an iso-8859-1 dict, that long string would be
passed to ispell -w
--
Agustin
