Barnum wrote:

> This is the XML input:
> <root>
>    <foo @name="1">a</foo>
>    <foo @name="1">b</foo>
>    <foo @name="1">a</foo>
>    <foo @name="2">c</foo>
>    <foo @name="2">c</foo>
>    <foo @name="2">d</foo>
> </root>

I guess

<root>
    <foo name="1">a</foo>
    <foo name="1">b</foo>
    <foo name="1">a</foo>
    <foo name="2">c</foo>
    <foo name="2">c</foo>
    <foo name="2">d</foo>
</root>

is more what you have.

> I expect to get only these nodes in fum:
>    <foo @name="1">a</foo>
>    <foo @name="1">b</foo>

You simply need to apply the predicate @name = '1' e.g.

   /root/foo[@name = '1'][not(. = preceding-sibling::foo[@name = '1'])]

or in that template

     <xsl:variable name="fie" select="$in/foo[@name='1']" />
     <xsl:variable name="fum" select="$fie[not(. = 
preceding-sibling::foo[@name = '1'])]" />

Defining a key is also easy:
   <xsl:key name="k1" match="foo[@name = '1']" use="."/>
then you can use Muenchian grouping with that key.

-- 

	Martin Honnen --- MVP XML
	http://msmvps.com/blogs/martin_honnen/