Re: [xsl] How to make this script faster

["Mathieu Malaterre" <mathieu.malaterre@xxxxxxxxx>]

On Nov 16, 2007 9:59 AM, Michael Kay <mike@xxxxxxxxxxxx> wrote:
> > Michael Kay wrote:
> > > But that's not a performance issue, and nor are most of the points
> > > Abel made; and I have to say I couldn't see anything at all
> > here that
> > > should cause performance problems.
> > >
> >
> > Not even the recursion?
>
> There's nothing intrinsically inefficient about recursion. As far as I could
> see, the process here was linear. Sibling recursion can lead to stack
> overflow if the number of siblings is high, but that doesn't make it slow. I
> looked hard at this to see if there was any non-linearity, but if there is,
> I couldn't find it.


Hi all, and thanks for comments. So here is the real test case:

Files are available at:

http://gdcm.svn.sourceforge.net/viewvc/*checkout*/gdcm/Sandbox/07_03pu.xml

http://gdcm.svn.sourceforge.net/viewvc/*checkout*/gdcm/Sandbox/dcmtk/extract_section.xsl

Then running saxon8 from command line:

$ time java -jar saxon8.jar 07_03pu.xml extract_section.xsl
...
java -jar saxon8.jar 07_03pu.xml extract_section.xsl  4.10s user 0.15s
system 38% cpu 11.123 total

And that's for only *one* section ! system is: debian stable,
$ java -version
java version "1.6.0_01"


Thanks !
-- 
Mathieu