Translation (was Re: [PlanetCCRMA] locale breaking redhat-config-services)

Fernando Pablo Lopez-Lezcano nando@ccrma.Stanford.EDU
Mon Sep 1 23:18:01 2003

> is there a translation project for Planet CCRMA? I'd be proud to translate
> to French (my native langage), so that I can help this project that is so useful
> for me.

Not yet. It would be _great_. Francois (of IRCAM) wanted to start one,
we talked about it while in the BYOL workshop in Prato. There are
practical problems on how to do it, but nothing that could not be

To give you an idea, currently the web pages are created with latex2html
from the original "source code" that is written in latex(*). It would be
fairly easy to translate that and have an "alternative" page in French
(or Spanish or whatever). The problem is how to keep it up to date and
how to id which parts of the document in English have changed. My idea
so far is to use some other latex to html document translator(**) that
supports latex's conditional execution structures. In that way the
source for the web pages would be one (or a set) of latex pages with the
different translations embedded in them, a suitably chosen variable
would determine in which language the document is rendered to. Splitting
into multiple pages and keeping them in cvs could make this work (and
maybe finding a way to tag in the source which paragraphs or sections
have changed since the last time the site was updated). 

-- Fernando

(*) Why latex?

Well, it does _not_ have to necessarily be latex but here are the things
anything being used would have to provide for me to be happy:

- one source file (or sources) in text format that can be edited by
normal text editors (I'm an emacs kind of guy but whatever is used
should be agnostic in relation to the editor to be used). Why? I don't
want to have to click around to edit stuff, I want it simple and based
on plain ascii text. The faster I can do things the better it is. 

- a "programming environment", at least in the sense that I could define
things like version numbers in variables so that I change them once and
have them updated everywhere. 

- obviously support for html constructs and some control on the
rendering of the text, inter-document links and things like are taken
for granted. 

- conditional execution or interpretation of the source. 

- the language does not have to be dynamic at all, I'm perfectly happy
with a static site that can be "recompiled" with a simple command. 

- probably many other things I'm forgetting... :-)

(**) I spent about a week doing research on this and so far the only
translator that looks promising is Hevea. It does have problems but I
managed to get a fairly decent looking Planet CCRMA clone built with it.
It does have support for conditionals and that is important, not only
for translation but also to be able to generate versions of the document
for all versions of Planet CCRMA - so that, for example, in the RedHat
8.0 version there are no references to anything that is related to
RedHat 9. So, in an ideal world a bunch of compiles would generate a
"Planet CCRMA for RedHat 7.3" in English, another in French, another for
8.0 in English and so on and so forth (and of course versions for SuSE
and Yellow Dog and all that, right?.. very easy :-)