Hi there,
I have to group information in the hypothetic format:
<item>
<prop1>x</prop1>
<prop2>y</prop2>
<prop3>z</prop3>
<lots-of-other-info-for-display/>
</item>
I need to group info in the items as follows:
prop1
prop2
prop3
(Thus - first by prop1, then prop2, then prop3)
To do this, I am using keys, but I am encountering very bad performance when
using Xalan. To improve the performance I have now produced only one key
using a concatenation of all properties to group the items:
<xsl:key name="per" match="//item" use="concat(prop1,concat(prop2,
prop3))"/>
After grouping I have to produce a fairly complex matrix (output as an html
table) for each item (since the grouped items are then combined and
displayed in a single row based on other item info). This also takes quite a
while, so I have resorted to taking a multi-pass approach by using the Xalan
nodeset function to first apply-imports to produce an xml document with the
grouped items and then apply templates to this nodeset to produce the html.
My question is this - is there some "rule-of-thumb" max size where the input
documents should rather be grouped or split by the processes producing them
rather than the stylesheet? I do not want to do this, since the grouping is
based purely on a presentation requirement and I do not want the application
or business layer to worry about these issues. With the above scenario,
using Xalan 2.3.1 on linux I seem to start having problems around the 2Meg
mark with 1500 items.
Any suggestions on alternative methods etc would be much appreciated.
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list