RE: [xsl] XSLT Architecture: Next Step

Subject: RE: [xsl] XSLT Architecture: Next Step
From: "Michael Kay" <mhk@xxxxxxxxx>
Date: Thu, 3 Jul 2003 21:00:40 +0100
> 
> From what I read on it and previously in other articles about 
> the subject, my first impression was that XSLT pourpose was 
> to perform the presentation of data in a browser or cel phone 
> or whatever (which is also well explained on Didier's 
> article). 

If you look at the intro to the XSLT 1.0 spec, you will find the
sentence:

XSLT is not intended as a completely general-purpose XML transformation
language.

If you understand anything about how committees work, you will recognize
that this sentence wouldn't be there unless some people thought it
should be a general-purpose transformation language, and they took a
vote on it, and decided that it shouldn't. But the vote changed nothing,
because by the time people noticed what was happening it had become a
general-purpose language anyway.

(I'm surmising, by the way. I wasn't on the WG at the time and I don't
know what the real story is; but I do have a good idea how these
discussions go).

In my view, XSLT needs to be "complete" in some sense, rather like the
relational model is complete, in its ability to transform XML trees. You
can't do the presentation job properly unless it is complete, and once
it is complete, you can do many other things with it too.

I also tend to the view that those facilities which can be seen as
special-purpose and oriented to presentation tasks, for example
attribute sets and the lang() function, turn out to be the features that
are least used.

Michael Kay



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


Current Thread