Re: i18n/gettext?

[Paul Eggert]

> Date: Sun, 9 Dec 2001 20:01:17 -0700 (MST)
> From: Richard Stallman <address@hidden>
> 
>     >  - What happens when a package is loaded or autoloaded?  How will its
>     >    catalog be read and added to the data base?
> 
>     If the package is part of the standard distribution, the catalog will
>     already be in the database.  If not, then see the first point above.
> 
> To do this, we need a way to indicate the proper catalog for any given
> string.  How would we do that?

Strings in the standard distribution wouldn't need any indication; they
would just be in the Emacs catalog.

Strings in other packages would need to do the equivalent of the
following C code:

  bindtextdomain (PACKAGE, LOCALEDIR);
  textdomain (PACKAGE);

where PACKAGE is the name of the package, and LOCALEDIR is where its
message file is stored.  We could establish a convention where
Emacs-related package locales are, but it might be simpler just to
stick with the standard GNU convention, e.g. for GNUS the localedir
would be /usr/local/share/locale/gnus, whereas for Emacs the localedir
would be /usr/local/share/locale/emacs.

>     For the English strings to be translated, we use ASCII.
> 
> Maybe we should have a rule that strings with non-ASCII characters
> are never translated automatically--if you want to translate them,
> you have to do it explicitly.

I don't quite follow what you mean by "translated automatically" here.
I've mostly been talking about the problem of automatically entering
these strings into a table that is given to human translation teams.
This table, and the resulting translations, are part of the Emacs
distribution (or the other package distribution, if it's some other
package).

My understanding is that, when Emacs itself is running, strings are
never translated "automatically"; the Emacs programmer must either
request an explicit translation, or must pass the strings to a
function that explicitly translates them.

But to get back to the issue of the message-ID character set: the GNU
gettext manual recommends that all message-IDs be ASCII strings,
because automatic character set conversion is not applied to them, and
because the strings should be useful as-is in case the translation
table doesn't contain an entry for them.  So I'm dubious about
suggesting to programmers that they try to translate strings that
contain non-ASCII characters.