At 10:41 AM 2/28/2003, you wrote:
On Fri, Feb 28, 2003 at 09:30:15AM -0500, Joseph Kesselman wrote:
> Just a quick observation: If you really prefer to think in terms of
> generating TeX rather than XSL-FO, IBM published a technology
> demonstration quite some time ago which consisted of a XML equivalent
> syntax for LaTeX and a postprocessor that would convert this into standard
> LaTeX markup. It might still be available from the alphaWorks website.
You mean TeXML? The problem with that is that it constrains the XML you
use in your original documents.
How so? Target it with a transform. Whatever primitives it gives you,
that's what you've got. Or make your own (and extend the back end): it's
(Ah I see you saw that. Now keep in mind, maybe a tag set isn't *in itself*
a "target format" or "source", but rather just better suited for one job or
XSL-FO doesn't make that assumption, and
I think that's a big argument in favor of XSL-FO.
Both are rendering languages. As such, they are not likely to be very good
for authoring or what-have-you data capture (though people have captured
data into HTML long enough, and that's arguably no better).
The constraint we've
been considering in this discussion has only been with the XSLT recipes:
should they transform arbitrary XML into XSL-FO or into something else?
XSLT stylesheets that truly work on *arbitrary* input (or are intended to)
tend to be diagnostic stylesheets of one kind or another; in my experience
most stylesheets targeting a renderer are pretty quickly customized to one
type of input (classically, one DOCTYPE). So much is probably obvious when
you think about it -- but it can be a subtler point than it seems.
Usually whatever you target from your transform, you'll make assumptions
about your input. So the question isn't just "what's the best target format
for my transform (most versatile, powerful etc.)" but also "what's the
easiest target to reuse over and over making more stylesheets for more
kinds of input?". That's at least the position a public "standard" has to
take. If you happen to know you'll never have anything but a very limited
and well-understood kind of input -- say XHTML? or name your tag set --
then you may well find targetting PostScript to be effective (I don't know
enough PS to say but I bet XHTML would be a bear). In which case -- you've
just invented another rendering format. Probably not much better for
authoring than the contenders!
Now if you truly had a stylesheet that could take *arbitrary* XML and do
smart rendering (whether it be via PS or not) ... *that* would be an
"Thus I make my own use of the telegraph, without consulting
the directors, like the sparrows, which I perceive use it
extensively for a perch." -- Thoreau
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list