Hi Duane,


Just thought I'd offer some advice that is similar to what Michael Kay
already mentioned but thought I might elaborate a bit on.

Given

> <root>
>    <apple/>
>    <pear/>
>    <orange/>
> </root>
>
> Given an xpath to orange (/root/orange) I want to reference the
> previous element which is a peer.

The peer is a pear, very true ;).  Back to the advice:

>I tried the following with no
> success:
>
>    /root/preceding::orange

Of course not,  this doesn't necessarily make a whole lot of sense.
Imagine every XPath to be similar to a directory path, only more
complex.

What does /root/../../ get you?  Something that doesn't really exist.
Now imagine a system that would allow you to get the "next"
directory/file in a list. so say root has four files: baseball,
cricket, dance, and Zamboni_Racing.  You could get the previous file
by previous::filename, or get all previous files by using previous::*.
 So you could do /root/dance/previous::baseball, or
/root/Zamboni_Racing/previous::*.  In XPath, this is preceding, but
that's actually more complicated. (preceding-sibiling is closer to
what I described here.)

Your XPath is looking for all the element nodes that are orange that
occur before root, not the previous node before orange.  Others have
mentioned the workaround for that.

One other observation, it can be complicated to use a lot of full
paths.  Not always, but usually it indicates some issues with having
too much pull.

Instead, you could have something like:

<xsl:template match="orange">
<xsl:for-each select="preceding-sibling::*[1]">
do some stuff
</xsl:for-each>
</xsl:template>

As always, I've made some simplications, but hopefully it might help
you read the XPaths a little better.

Jon Gorman