troubleshooting XSLT performance

Hi,

I am using .Net 2.0 to run an XSLT process.

It basically takes a word XML document at it's input, plus a separate XML
file opened using the document() function and applies changes to the word
document on the basis of the secondary file.

A customer has reported that for one document, the process takes about an
hour to run (the word doc is 4mb and the file of changes is about 100k)
whereas for other similarly sized files it normally takes a few seconds.

we have been unable to reproduce the problem in the development environment.
clearly the problem is data dependent but the customer's data is
confidential so they can't send the files to us

is there any way I can get any kind of diagnostics to find out what is
slowing it down?

TIA

Andy

re: troubleshooting XSLT performance

Andy Fish wrote:
There is a profiler
http://blogs.msdn.com/xmlteam/archiv...r-vs-2008.aspx


--

Martin Honnen --- MVP XML
http://JavaScript.FAQTs.com/

re: troubleshooting XSLT performance

"Martin Honnen" <mahotrash@yahoo.dewrote in message
news:OjjD9RNGJHA.5780@TK2MSFTNGP03.phx.gbl...hmm thanks, I didn't know about that one and it will definitely come in
handy in our development environment

unfortunately at the moment I'm stuck with the customer's production
environment. I can deliver him a special build with instrumentation in but I
can't install visual studio on his production box.

re: troubleshooting XSLT performance

* Andy Fish wrote in microsoft.public.dotnet.xml:You can send them an XSLT that removes all confidential data from the
document, like removing attributes not used in the transformation, use
replacement text for text nodes, and so on. They could then send you
the remaining skeleton. How easy this would be depends on the formats
in question of course.
--
Björn Höhrmann · mailto:bjoern@hoehrmann.de · http://bjoern.hoehrmann.de

re: troubleshooting XSLT performance

"Bjoern Hoehrmann" <bjoern@hoehrmann.dewrote in message
news:ci82d45b80st2rmu9mnbsirbeptbg25icl@hive.bjoer n.hoehrmann.de...hey that's a neat idea. the document is Word XML so all the words (and only
the words) will be within <w:telements. I think they probably have images
and diagrams but we could work around that...

at the moment i'm working on changing the XSLT to add logging using a custom
script block to call log4net. hopefully they can send me the resulting log
file and i should be able to figure out from that what's happening. if that
doesn't work, I think yours will be my next stop

thanks

Andy