Re: [xsl] follow up re: key execution speed

[Jeni Tennison <mail@xxxxxxxxxxxxxxxx>]

Hi Mike,

> The trick is this: while I have been advised that keys are speed for
> any lookup that will be performed more than once in a transform, my
> situation is a little more complicated. I only do this lookup once
> per transform, but I perform this transform many times (thousands!)
> on different source documents. I was hoping to get the speed benefit
> of the key across these multiple transformations.

If the multiple transformations happen in the same process, then
perhaps you can adjust your XSLT to pull in the thousands of source
documents sequentially, and perform the transformation of each. That
way the key is produced once, used multiple times, and you'll get the
benefit of the key.

Otherwise, there is no point in using the key, and I imagine it makes
performance worse because of the memory usage. Instead you may as well
search for the relevant ZIP and DEALER information using an XPath.

However, the real problem is that parsing big documents takes time.
But there are a couple of other things that might speed things up.
First: rather than having big Zips.xml and Dealers.xml document, split
them down into multiple XML documents (using XSLT if you like),
indexed by their values. So for example, Zip.10001.xml being:

<ZIP rc="2" />

You can then quickly access the value that you want using:

  document(concat('Zip.', $ZIP, '.xml'))/ZIP/@rc

The parse of Zip.10001.xml won't take long because it's so small.

Alternatively, if you don't want to have to manage all those files,
put all the information into a database and use a specialist
EntityResolver (or whatever it's called) so that when you use
document('') with a 'zip' protocol, it checks the database for
something with that zip and returns a document that looks like the
above. (Or create an extension function in Java to retrieve the Zips
from the database.)

Another thing that you may be able to do is to pre-parse the Zips.xml
and Dealers.xml documents, hold them as DOM Document objects and
either pass them in as parameters to the stylesheet or again write an
EntityResolver that proffers those DOM Document objects on demand.
That way they only get passed once. Again, the key will make no
difference since the lookup only gets done once per transformation,
but you'll have got around the big barrier of parsing the documents
each time.

Cheers,

Jeni

---
Jeni Tennison
http://www.jenitennison.com/


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list