I have been using the following xpath selection in xpath 1.0
<xsl:variable name="id"><xsl:value-of select="./@id" /></xsl:variable>
preceding::exm:messageFlow[@sourceRef = $id]/@targetRef
This worked very well and did exactly what I wanted.
However I started using XSLT 2.0 in combination with xPath 2.0 and this does not work anymore
(Using the Altova XML processor)
I get the following error:
Kann nicht mit Ziel-Typ besetzt werden – Aktuelles Element ist ‘sid-B3FD7EE5-043
3-4939-A69F-E74B30FDEB1C’ vom Typ xs:untypedAtomic, Typ xs:QName erwartet – =
which translates roughly to:
Cannot be set to target-type – Current element is ‘sid-B3FD7EE5-043
3-4939-A69F-E74B30FDEB1C’ of type xs:untypedAtomic, type xs:QName
expected –
Previously I had a similar problem with this selection:
following::exm:*[exm:incoming = $out] | preceding::exm:*[exm:incoming = $out]
Which again worked in xpath 1.0, but returned a similar error in xPath 2.0
After changing it to
following::exm:*[exm:incoming/text() = $out] | preceding::exm:*[exm:incoming/text() = $out]
It worked in xPath 2.0 as well. I did try doing something similar with the attributes, but it did not work
A small extract of the corresponding XML:
<messageFlow id="sid-80B618A4-E6BF-4438-AF5D-5111AD308FE6" name="" sourceRef="sid-B3FD7EE5-0433-4939-A69F-E74B30FDEB1C" targetRef="sid-6EB2DB76-CC19-48AD-A073-D37C7489F211"/>
<task completionQuantity="1" id="sid-B3FD7EE5-0433-4939-A69F-E74B30FDEB1C" isForCompensation="false" name="call service" startQuantity="1">
<incoming>sid-2B2BA651-B5BA-4195-9B5B-E6855B1138F4</incoming>
<incoming>sid-DA4B86E8-C3F2-497C-8C0D-218E95CE9FD1</incoming>
<outgoing>sid-008948DE-BA59-4897-AC37-2C3AA63DCD82</outgoing>
</task>
I wonder if you have a schema that declares the @id attribute as being of type QName?
With this variable declaration:
you are creating a copy of the @id attribute in a new result tree fragment. This is quite unnecessary; your needs would almost certainly be better served by making the variable be simply a reference to the existing attribute, thus:
By making the copy, you are not only writing unnecessary code and incurring unnecessary run-time cost (building a new tree is an expensive operation), but you are also losing the type information. If my conjecture is correct that @id is of type xs:QName, then the variable $id after atomization will be of type xs:untypedAtomic, and comparing it to an xs:QName is likely to fail with a message similar to the one cited.