Jan Egil Hagen <janha@ifi.uio.no> writes:

> There should be a variable gnus-babel-languages so that it will be
> possible to do something like:
> 
> (setq gnus-babel-languages
>       '(("^fr" "French -> English" t)
>         ("^de" "German -> English" nil)))
> 
> The third element in each list says whether article-babel should
> *always* translate, or just default to this value if called
> interactively.

That's a good idea.  I've now added a `gnus-treat-translate' variable, 
but I'm wondering a bit about a slightly more general thing, and
that's the controlling-what-groups-a-treatment-function-is-run-in
thing.  It's more obviously an important thing with this function than 
with, say, gnus-treat-display-smileys, but perhaps there should be a
general mechanism.

Should the treatment variables also allow, say, a list of regexp to
match groups where they should be used?  Gee, that seems like a pretty 
simple way to solve that problem.  :-)

Anyway, back to the translation thing -- would it make more sense to
specify what the preferred language one reads in is (say, "en"), and
then have a mechanism that says what language groups are in (say, a
whopping big alist of group regexps to languages (anyone want to write 
that alist?)), and then Bob is your mother's brother?

Like:

(setq gnus-preferred-language "en")
(setq gnus-treat-translate '("^fr" "^de"))

This would then mean: «In the "fr" and "de" hierarchies, translate to
English».  Gnus will know what language to translate from based on
that whopping big alist and possible Content-Language tags in the
messages themselves.

(If the preferred language/original language pair is untranslatable,
then Gnus won't do anything.)

Whadday'allthink?

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen