Template mechanisms...

Richard A. O'Keefe ok at cs.otago.ac.nz
Fri Dec 13 03:40:32 UTC 2002


goran.hultgren at bluefish.se wrote:
	Me too. I have a collegue that likes XSL but I am trying to argue that
	webdesigners don't use it and don't know how to. Not their tools either,
	that's my guess of course.
	
	I have also heard that performance is tricky.
	
I've tried several of the better known XSLT implementations.
It's important to realise that

(a) XSLT is layered on top of XPath, which *may* possibly have some kind
    of theoretical basis allowing efficieny compilation, but if so,
    no-one seems to talk about it
(b) XPath and XSLT are both moving targets; the W3C committees seem to
    be extremely fond of taking an overly complex and poorly defined
    interface and blowing it out further.
(c) Given the historic basis of DSSSL in (pure declarative) Scheme,
    and the revulsion of the W3C people against non-XML syntaxes,
    it's rather sad that Scheme code for manipulating XML is *amazingly*
    smaller and simpler than XSLT.  Faster too.
(d) Most of the better known XSLT implementations take a further hit
    by being written in Java.  (There's lots of fun like the definition
    of nth-character-of-a-string being *different* in XPath (hence XSLT)
    and Java (hence the DOM), but that's W3C for you.)

Given that context, the only XSLT processor I've come across which isn't
as slow as a snail crawling over concrete is libxslt, and the last time
I played with that the documentation made Squeak look good.




More information about the Squeak-dev mailing list