Rick --
I'm not sure if this was ever resolved to your satisfaction. The format
that you had been using:
<xsl:variable name="Values"><xsl:copy-of
select="/Rowsets/Rowset/Row"/></xsl:variable>
<xsl:variable name="SPCResults" select="java:LHSPC.XIR($Values)"/>
should have passed a single node to your extension function. This is
because it creates the Values variable to be of type
result-tree-fragment which is defined to have a single root node in the
XSLT Recommendation. The current implementation is correct and the
previous implementation was not correct if it passed all of the nodes to
the extension function.
Actually, this may be just changing the way that we're mapping RTFs to
java NodeLists. We are now mapping them so that an RTF is passed as a
java NodeList consisting of a single root node. In the past, we may
have mapped an RTF to a java NodeList by creating a NodeList consisting
of all the child nodes of the RTF root. In my view, the current mapping
is more logical since it more accurately reflects the XSLT structure.
Gary
> -----Original Message-----
> From: Rick Bullotta [mailto:[EMAIL PROTECTED]]
> Sent: Wednesday, September 19, 2001 7:23 AM
> To: [EMAIL PROTECTED]
> Subject: RE: Extension Function Problem/Incompatible
> DocumentFragmentImpl implementations?
>
>
> The core issue behind this bug relates to some type of
> internal difference between the results of a select and a
> copy-of relative to variable assignment. The following works
> with our extension function, and, as expected passed a
> properly populated NodeList to the extension function:
>
> <xsl:variable name="Values" select="/Rowsets/Rowset/Row"/>
> <xsl:variable name="SPCResults" select="java:LHSPC.XIR($Values)"/>
>
> The following, however, did not work as expected (although it
> worked with versions prior to D10):
>
> <xsl:variable name="Values"><xsl:copy-of
> select="/Rowsets/Rowset/Row"/></xsl:variable>
> <xsl:variable name="SPCResults" select="java:LHSPC.XIR($Values)"/>
>
> In the latter case, a NodeList contain a single Document node
> appears to be passed to the extension function.
>
> Hope this helps pinpoint the problem!
>
> Regards,
>
> Rick Bullotta
> CTO
> Lighthammer Software
