Re: [xsl] XSLTPROC performance

[Tony Graham <Tony.Graham@xxxxxxxxxxxxxxxxxxxxxx>]

On Thu, Dec 20 2007 09:38:47 +0000, gr@xxxxxxxxxxxxxxxxxxxxxx wrote:
...
> The problem is that we made some benchmark (not very extensive benchmark 
> indeed) and seems that the XML::LibXSLT solution is 20% -50% slower in 
> execution than the XSLTPROC solution. Can anyone confirm or disconfirm this 
> results? I've surf your mailing list archivie and the net but I can't found 
> comparative benchlmark.

XML::LibXSLT is libxslt plus a layer of interface code between libxslt
and the interpreted Perl program.  It's not surprising that it's slower
than xsltproc.

> If it is true, than I have to extend XSLTPROC (performance is the highest 
> requirement), is there any tutorial how to do this (I've already red the one 
> in the homepage od LIBXSLT and the one that comes with the source 
> distribution).

I haven't seen any other tutorial, but you can look at the source code
for the tests in the libxslt source code distribution to see how they do
things.

One thing that's undocumented on the libxslt web site is that recent
versions support extensions as 'plugins' in a 'libxslt-plugins'
directory as an alternative to the previous requirement to make your own
xsltproc-variant that linked to your own extension functions.

Regards,


Tony Graham.
======================================================================
Tony.Graham@xxxxxxxxxxxxxxxxxxxxxx   http://www.menteithconsulting.com

Menteith Consulting Ltd             Registered in Ireland - No. 428599
Registered Office: 13 Kelly's Bay Beach, Skerries, Co. Dublin, Ireland
----------------------------------------------------------------------
Menteith Consulting -- Understanding how markup works
======================================================================