[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: XLink transformations

• From: Kay Michael <Michael.Kay@i...>
• To: 'Steve Boyce' <SteveB@h...>, "'xml-dev@x...'" <xml-dev@x...>
• Date: Tue, 18 Jul 2000 14:13:33 +0100

> And this relates very obviously to a (mis)understanding I have about
> XSL/XSLT.  Don't these "secretly" presuppose that they are 
> mapping from one
> schema to another schema?  I mean, if I write a stylesheet, 
> in reality I
> have in mind a source and destination schema.  Shouldn't these be made
> explicit?

Some stylesheets can be written to process XML usefully regardless of its
schema, for example a simple "pretty-printing" stylesheet. You are right,
though, that in general a stylesheet will make certain assumptions about the
structure of its input, and it should be possible to describe these
assumptions in a schema. And if this were done, it would arguably be
possible to optimise the stylesheet execution by using this extra knowledge.
(A simple example is that //item need only search the tree to a certain
depth).

It's also true that the stylesheet writer will have a schema for the result
document in mind. Whether there is anything that can be usefully done with
this, other than validating the output against it, I'm not sure. Some
stylesheet errors could be detected at compile time from knowledge of the
result schema, but probably not enough of them to be useful. And I can't
think of any way of using this information for optimisation.

Mike Kay 

• Follow-Ups:
  • Re: XLink transformations
    • From: Martin Bryan <mtbryan@s...>

• Prev by Date: Re: List member issues
• Next by Date: Re: XLink transformations
• Previous by thread: RE: XLink transformations
• Next by thread: Re: XLink transformations
• Index(es):
  • Date
  • Thread