Re: Emacs i18n

[Eli Zaretskii <eliz@gnu.org>]

> From: Jean-Christophe Helary <brandelune@gmail.com>
> Date: Fri, 8 Mar 2019 10:48:46 +0900
> 
> I think what really needs to be discussed is:
> 
> • which strings do we extract
> • how to rewrite the mix of code and strings
> • how to extract the resulting strings
> • how to process the translations for display in emacs

Extraction is just a technicality, it can be done in either of several
possible ways.  We could use xgettext, or we could use a modification
of make-docfile (the latter is probably a must for collecting do
strings from C sources), or we could use po4a or something similar.
As long as the catalogs are PO files, we could even use a mix of
tools, if, for example, some of the tools is more convenient for Lisp,
but not for C.

And I don't understand what problems you see in the last item: what
should be done there other than display the translated string with
'message' or insert it into the *Help* buffer?

So I think you are bothered by stuff that is largely non-issues.  The
most important issues IMO are different: (a) what methodology of
extracting/marking translatable strings to choose so that this job
doesn't become infeasible; and (b) how to arrange the message catalogs
so that they will be easy to maintain and update, given the modular
nature of Emacs.  I think we should also take a better look at how the
built-in help facilities generate documentation and other displayable
strings from symbol names.  Macros such as define-minor-mode should
also be scrutinized to see if there are some special problems there.

Once this is done, the methodology decided, and the necessary tools
are available, the rest is just more or less mechanical work to
convert more and more parts of Emacs.