David Carlisle wrote:
I'd be surprised if either of those made any difference.
Both examples I gave were guesses (I wrote "I suspect"), that's why I
need a profiler.
If I knew all the bottlenecks, I would not need a profiler ;)
// is expensive in select attributes but in a match it's just doing
nothing.
I see.
similarly a processor has to cache the result of document() (almost
certainly, to get the node identity semantics right) so explictly
saving it in a variable is unlikely to make much difference (but does
make the code clearer)
I heard that the second version would be faster. But that's why I want a
profiler: I can rely on its statements, instead of relying on rumors in
books or lists.
Instead of guessing which changes improve performance, then do trial and
error for hours in the wrong places, I need a tool telling me where the
bottlenecks are, so that I can do the trial and error where it matters,
isntead of where it doesn't.
Tobi
--
http://www.pinkjuice.com/
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list