> Note that Liam wrote "using explicit typing on variables".
> As I read it, he claims that XSLT 2.0 can be executed more efficiently
> than XSLT 1.0 when the XSLT code is statically typed using XML Schema,
> likely because runtime optimisation of untyped
In 99% of cases a couple of
algorithm fixes will do much better than tons of all that low-level optimizations.
So I don't believe in all that "typed variables" stuff, especially relatively
to modern processors. Yeah, it can(!) sometimes(!) give you +10% speed, but hey!
I can get +700% speed on a cheap multicore Opteron by simple adding threads. And then I'll
get 1000% over that by implementing a VERY basic caching. And someone is going
to make me play all that "typed/untyped" games now? Which is particulary
funny with Java, btw. I don’t buy it, wait 3 years or more and then retry.
XSLT1 have some limitations, but I have a bunch of inexpensive HTML-coders working with it, and they are as hapy as my programers are, working with pure XML. XSLT2 will make our life a bit easier, perhaps, but it will change nothing in general. So I'm currently focused on the core task: make libxslt as fast, as it should be from the start.