On Sun, Mar 06, 2005 at 04:45:28PM +0100, Jordi Vilalta wrote:
 >Damn, you people are going to want a working HTML po4a
 >module soon, aren't you? :)
 I've done some small tests and it seems that the Xml module is currently 
 enough for our needs :) 
Yes, that's the conclusion I more or less reached yesterday:
HTML is too messy, and it's better to mandate the code to
comply to XHTML, in which case the XML module seems like
it'll do the job nicely (although I haven't finished that,
and considering you know it very well you'll probably be
done before me...)
 >We've got little flags on our site (
www.naryves.com), which
 >the CGI will only display if the current page is actually
 >translated. However:
 It's okay if there's a small amount of languages, but it isn't well 
 scalable. 
That's true; I suppose what is done on 
www.debian.org or
www.gnu.org is the simplest way.
[Language-accepted]
 I haven't tried it, I've directly gone to (successfully)
detect it in php. 
 I prefer having the control over these variables, and this way we don't 
 need duplicate files (index.html.*). 
I'm confused. Are you running po4a to get the translation of
the page for each request, using a php script? If so, that
sounds vastly under-efficient.
What I do is that for each master file (foobar.shtml.en...
I assume there is only one master language) you generate as
many PO as translations (foobar.fr.po, foobar.de.po and so
on) then a script goes through the PO files and generates
all the translations (foobar.shtml.fr, foobard.shtml.de).
I don't really see how you can avoid duplicating the files.
Y.