Hi,

i'm looking for a way to speed up a xslt transformation.

Input.xml looks something like:

<Input>
    <E i="1" a1="1" a2="2" />
    <E i="2" a1="1" a2="2" />
</Input>

the desired output FILES are:

<Output1>
    <E i="1" a1="1" />
    <E i="2" a1="1" />
</Output1>

AND

<Output2>
    <E i="1" a2="2" />
    <E i="2" a2="2" />
</Output2>

(The files contain a different SET of attributes)

Well, what i already have is an parametrised xslt which generates those "excerpts".
I just thought that using a XSLT for this kind of job
(the real input files are about 40Kb and have to be transformed several times per second)
is probably not the most efficient way to do it.

I'm looking for creative ways to eighter reformulate the xslt
or a completly different approach to splitt those files (regexp?)

What i'm using right now is: C#. System.Xml, xsltpathDocument etc.
The transformation is already cached.

My thought is that the input xml file structure is fixed,
so maybe i could somehow just fill in the new values
and use some other mechanism (not xslt) to generate those excerpt files.

The problem with c# xslt is, that it allocates dynamicly alot of memory to process
the files, so even though the xslt itself is cached and preprocessed
the processing of a new input file causes the system to allocate memory
from the gc, which i'd like to avoid as this is an identified slowdown
in the whole process.

Thanks for ANY ideas and thoughts about this

cheerio
Radek