RE: [xslt] Benchmarking

["Robert Koberg" <rob koberg com>]

> I was speaking of the trade-off decisions made while developing the
> processor (note the past tense). At the time standard compliancy was
> more important than speed, pattern matching speed was more important
> than recursion speed, etc. We had a lot of ground to cover, and
> recursion was low-priority. Things may have changed since then, and
> I am certain that Daniel will appreciate any contributions that
> optimizes any part of libxslt, including recursion.

I am just a dumb user. I wanted to point out that recursion is very
important and find out why you guys felt it wasn't. I understand what you
are saying and what you have produced is my current choice. Given that you
did not create this with performance in mind it is a fantastic job!

Michael Kay has a section header in his Pro.XSLT book (wrox 1st ed.) on page
551 saying "Don't Iterate, Recurse". Many people use this (and the 2cnd ed.)
as THE XSLT bible.  In that section he goes on to say:
<paraphrase>
Is recursion expensive? No, a smart complier should be able to produce the
same code from a recursive operation as an iterative one.
</paraphrase>

Note: I would rather have bugs fixed than improving recursive procedure
speed (which is still faster than the java implementations I used before
libxslt).

best,
-Rob