Re: Benefits of using xsl:key

2009/11/3 Jesper Tverskov <jesper.tverskov@xxxxxxxxx>:
> On Tue, Nov 3, 2009 at 9:03 AM, David Carlisle <davidc@xxxxxxxxx> wrote:
>>However a typical
>> muenchian grouping example can easily be orders of magnitude faster and
>> so measurable on really quite small documents as you are avoiding
>> an order  n^2 along the sibling axis looking for duplicates.
>
> What about XSLT 2.0 then, where we use xsl:for-each-group?
>
> A set of XSLT 2.0 test files demonstrating a dramatic difference would
> really be much appreciated?

Create a large index file along the lines of:

<entry id="nnn1" value="vvv1"/>
<entry id="nnn2" value="vvv2"/>

then in another stylesheet do multiple lookups into that file:

$entry-doc//entry[@id eq $id]/@value

then use a key instead:

<xsl:key name="entry-by-id" match="entry" use="@id"/>

with

key('entry-by-id', $id, $entry-doc)/@value

Provided the processor doesn't optimize // (such as the paid-for
version of Saxon) then you should see a measurable difference between
the former and the latter.


--
Andrew Welch
http://andrewjwelch.com
Kernow: http://kernowforsaxon.sf.net/