Re: ispell-aspell-find-dictionary: Opening input file: no such file or directory

[Magnus Henoch]

Eugene Vlasov <address@hidden> writes:

> After this:
>
> 2005-09-24  Magnus Henoch  <address@hidden>
>
>         * textmodes/ispell.el (ispell-maybe-find-aspell-dictionaries):
>         New function, code extracted from ispell-valid-dictionary-list.
>         (ispell-valid-dictionary-list, ispell-accept-buffer-local-defs):
>         Call it.
>
> spell checking on my system broken:
> $ emacs -q
> M-x flyspell-mode <RET>
>
> Loading flyspell...done
> Loading regexp-opt...done
> ispell-aspell-find-dictionary: Opening input file: no such file or directory, 
> /usr/share/aspell/ru.dat
>
> M-x ispell-buffer <RET>
>
> ispell-aspell-find-dictionary: Opening input file: no such file or directory, 
> /usr/share/aspell/ru.dat
>
> $ ls /usr/share/aspell/ru*
>
> /usr/share/aspell/ru-lebedev.dat
> /usr/share/aspell/ru-lebedev_phonet.dat
> /usr/share/aspell/ru-rk.dat
> /usr/share/aspell/ru-rk_phonet.dat
>
> $ aspell dicts
>
> en
> en_CA
> en_CA-w-accents
> en_CA-wo-accents
> en_GB
> en_GB-w-accents
> en_GB-wo-accents
> en_US
> en_US-w-accents
> en_US-wo-accents
> ru
> ru-lebedev
> ru-lebedev-ie
> ru-lebedev-io
> ru-rk

Sorry about the delay...

I've investigated this, and I'm not sure whether the bug is in
ispell.el, in the dictionaries used, or in aspell.

All dictionaries specify a .dat file (near the beginning of the binary
file).  The aspell manual says that this is named `LANG.dat', but the
ru-{lebedev,rk} dictionaries use ru-{lebedev,rk}.dat instead.  I'm
reluctant to parse the dictionary file from ispell.el, and "aspell -d
ru-lebedev config lang" gives "<unknown>" - making that return the
name of the .dat file minus .dat would be one way of fixing the problem.

The particular RPMs of these dictionaries that I found install
symlinks in the alternatives system, so that using the "ru" dictionary
will select one of them.  Given that, they could simply reference
ru.dat and get the right file, but that might not be in the in the
interest of the upstream maintainer.

On the third hand, ispell.el could try to find the real name of the
used dictionary (beyond symlinks), and try DICT.dat if LANG.dat
doesn't exist.  That would fix at least this particular problem.

So, I'm asking for advice: which of these should be fixed?  Or is
there another possible way?

Magnus