I'm working on a similar exercise for a client at the moment, and of course
I'm doing it all using Saxon. Saxon is issued with a profiling trace
listener and a stylesheet to analyze the data, but you will want to
customize these to get anything useful out of them. I find that the key
thing is to model the problem on a different scale, or a range of varying
scales, so that you can study the scaleability, and do experiments that take
seconds rather than hours to complete.
The stylesheet in this case looks monstrous, can you explain how it came to
be so big? The data size is within reasonable limits, Saxon can usually
handle this with about half a gig of memory. There's no reason to look for
non-Java solutions.
I tend to find that a detailed code inspection reveals the things that need
changing, but I've never had to inspect a 500K stylesheet.
Stylus Studio has a debugger that currently works with Saxon 6.5.3: watch
this space for further announcements.
Michael Kay
http://www.saxonica.com/
-----Original Message-----
From: Dusan Zatkovsky [mailto:zatkovsky(_at_)printsoft(_dot_)cz]
Sent: 10 September 2004 10:58
To: xslt mailinglist
Subject: [xsl] linux xslt debugger/ profiler
Hi all.
Has anybody know some good xslt profiler / debugger for linux,
especially NOT written in java? I have try catchXSL and editix, but
first ends with JAVA:Out of memory and second with unknown error
message. I am converting big (=20-60MB) xml files with big (=500k)
stylesheet.
I want to find points in my stylesheet which may be optimalised to
better processing speed.
Thank you.
--
Dusan Zatkovsky
--+------------------------------------------------------------------
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list
To unsubscribe, go to: http://lists.mulberrytech.com/xsl-list/
or e-mail:
<mailto:xsl-list-unsubscribe(_at_)lists(_dot_)mulberrytech(_dot_)com>
--+--