RE: Optimisation (was Remove duplicates from a list)

[Kay Michael <Michael.Kay@xxxxxxx>]

> This is entirely conjecture, but I predict that optimization 
> will follow a similar path as programming language compilers.

Yes, though I suspect SQL is a better analogy, because we're dealing with a
more declarative language. And it took about 10 years for the theory of SQL
optimisation to become refined.

There's certainly a lot of scope for optimising single XPath expressions,
and in some simple ways products already do. Unfortinately there are lots of
features of the language that make it difficult, for example the fact that
variables are untyped, and the very strange behaviour of the boolean
operators (like the fact that @A=@A can be false).

Optimising some other constructs - notably xsl:number - is going to be
really hard work, given the way they are defined.

One thing that would certainly make optimisation more feasible would be if
the stylesheet was bound to a DTD or schema and constrained only to process
documents that were valid against that schema. Unfortunately that would be a
change to the language.

Mike Kay


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