On Wed, Dec 13, 2006 at 06:00:16PM +0100, Danai SAE-HAN wrote:
> From: Ming Hua <firstname.lastname@example.org>
> > > From: Ming Hua <email@example.com>
> > > I'm convinced that the angle from which we should make up such
> > > typesetting rules should be from a LaTeX perspective. It is after all
> > > the engine which finally produces the PDF documents. The (X)HTML
> > > language is still too much in its infancy in order to distill
> > > typographical rules from it.
> > >
> > > I have a few suggestions.
> > [specific LaTeX typesetting suggestions]
> > For the case in point, Debian release notes is written in Debian SGML,
> > and both the HTML and LaTeX are auto-generated. So I suppose using the
> > specific LaTeX CJK typesetting syntax is not really practical here.
> LaTeX output is indeed auto-generated, but I think SGML is flexible
> enough to allow CJK breakable spaces. Don't we just have to create
> new entities? My way of thinking is this: LaTeX produces the most
> sophisticated but also the most technically challenging output. If
> SGML can manage to handle CJK LaTeX, then the other formats should be
> easy to convert to.
Sure, we can generate correct CJK LaTeX from SGML, but that probably
needs work (I don't really know). My point is, as SGML is also the
source of the HTML, we can't put, for example, the CJK space "~" in the
SGML file, because although LaTeX can handle it, HTML can't, and
therefore we'll end up with a bunch of literal "~" characters in the
And if you are suggesting using entities like "&cjkspace;" in the SGML,
then I'll probably object. It's too cumbersome and counter-intuitive
for spaces that are so frequently used.