Stylesheet optimisation

> Ouups, forgot the links to java lobby post :
>
http://www.javalobby.org/servlet/News?action=displayStories&xsl=comment.xsl&
format=full&id=510100000000125

Interesting article. Having experimented with various ways of avoiding
building the full tree, I'm sure many interesting transformations can be
done this way, but the language to achieve them won't be XSLT. XSLT gives
you direct read access to the input document and append access to the output
tree. Other possible approaches are to have serial read access to the input
and direct access to the output, or to have serial access to both and direct
access to your own intermediate working storage, or to have direct update
access to the input tree. These different processing models would result, I
think, in transformation languages that are very different from XSLT. 

I'm reminded of the days when ICL was a mainframe manufacturer and we
produced a rather clever 4GL for report writing and file transformation: it
was based on a Jackson Structured Design model in which the processing had
to be described as a sequence of phases each of which was either a serial
pass or a sort. It worked very well for processing gigabyte-sized files on
machines with megabyte-sized memory, and I'm sure the same idea would work
equally well for transforming large XML files, but it wouldn't be XSLT.

Mike Kay


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list