Lists Home |
Date Index |
On Fri, Oct 04, 2002 at 01:08:57PM -0400, Simon St.Laurent wrote:
> XML directly on the Web seems to have fallen victim to the notion that
> XML needed a transformative style approach, missing the easy opportunity
> that CSS provided for document display and requiring people to use XSLT.
> That notion has also provided Microsoft with plenty of cover for their
> (non-)approach to XML in the browser, which may have successfully kept
> XML off the ordinary Web.
Yes, but as you say, XML needs a transformative approach -- a way of
turning an XML format into a Java Class (e.g. WSDL -> SOAP proxy).
You can't do that if your only concern is document display, or your only
tool is CSS. Remember that XML was meant to live outside of the web, too.
> XSL ran into several challenges, though it seemed to keep going its own
> way regardless.
Looking at its current incarnation, it's difficult to see that XSL has
successfully overcome the biggest challenges it faced. In the early
days, ISTR a Scheme-like syntax. It was similar in its syntax, but
vastly different in its semantics than DSSSL. This would have been bad
because (1) lots of parens scare people off, and (2) it was
significantly different from the DSSSL model. Then again, it may have
been a straw man proposal...
The other issue that XSL overcame was the XML-based "rule" syntax.
XPath is *much* easier than that unfinished approach. (All I remember
is that it was an unfinished straw man, with lots of "this isn't quite
finished yet" edge cases.)
> The XSL community seemed, from the perspective of a CSS
> user, to have little interest in and much contempt for the notion of
> formatting through annotation generally and CSS particularly. The
> "Formatting Objects Considered Harmful" argument may not have bothered
> people who considered FOs a necessary result of a transformation, but it
> certainly troubled those of us who had hoped XML would encourage the
> sharing of computer-interpretable information.
ISTR the "FO Considerd Harmful" argument as something completely
different. Michael Leventhal made the case that FO was a very
inappropriate document format for the browser. Furthermore, anything
that could be done with FO would be done with less effort (and less
I think the problem with Michael's argument was that FO wasn't
intended to be rendered in a browser like HTML. Furthermore, FO
was expected to be the output of some program (like an XSL
transformation), rather than hand-coded. FO was designed specifically
to handle print media, not the browser. Years later, it's obvious
that these "problems" with FO were just a misunderstanding of the
intent. It was never put forth as a verbose replacement for HTML.
Overall, I think XSLT is quite successful. I'd go so far as to say that
XSLT's importance is second only to XML (then again, I'm quite biased).
One unexpected advantage with XSLT is that it fits well with the
pipeline model of serial XML processing. Each individual transformation
can be a very small, easy to write and easy to understand stylesheet.
Technically, it's possible to do the same thing with a stack of
SAX handlers or DOM transformations, but I'd much rather read three
small stylesheets than deal with the syntactic sugar to make a SAX
handler in Java/Perl/Python/etc.
> On the bright side, XML has certainly found use beyond traditional Web
> development, and XSLT has found plenty of use in styling (mostly
> generating HTML or HTML+CSS, ironically enough) and in many other
> transformation situations.
I'm not so sure it's ironic, considering that one of the most popular
DSSSL transformations was DocBook -> HTML/HTML+CSS. (Thanks, Norm!)