XSLT benchmarks

[Rick Jelliffe <ricko@allette.com.au>]

From: Eric van der Vlist <vdv@dyomedea.com>

> On the other hand, a new benchmark [4] published on XML.com shows that
> XT is, by far, the fastest open source XSLT processor (for any
> implementation language) and the only processor that seems to be able to
> compete with Microsoft's implementation.

Of the test results,
  * XT was the fastest for 7 of the tests
  * LibXSLT was fastest for 10
  * MSXSLT was fastest for 10
  * XSLTC was fastest for 13

This suggests to me that, thought MSXLT seems the least risk to start, for a
deployed real-time system you still need to benchmark.

It would be nice to have the results all in a single line graph so we could
judge the total effect more.

(It seems to me that the Python performance results need a little
interpretation: unlike Java it seems that they were done through a
command-line interface. If python has an initialization overhead, that will
be factored into the results.  Also, I am not clear on the XSLTC figures
whether the number includes compilation or not.   I bet all the data was
ASCII too.)

Cheers
Rick Jelliffe