Re: grouping after filtering plus a little more...

Hi joerg,
thanks for your reply. I have try the solution you've provided however, it seems to give me the same results as before (i.e. some data can not be display)... i had a look at the source file, it did not displays some result because the processor (i am using xalan-c 1.3) already spotted the data beforehand. (actually i had just spotted a mistake in my example ... i was suppose to group the type only, given that author='author2'. So my key should be <xsl:key name="books" match="book" use="type"/>
)

in my example ...
<book>
<author>
<name>author1</name>
<gender>m</gender>
</author>
<name>x box</name>
<type>sc-fi</type>
</book>
<book>
<author>
<name>author2</name>
<gender>f</gender>
</author> <name>elf story</name>
<type>adventure</type>
</book>
<book>
<author>
<name>author1</name>
<gender>m</gender>
</author>
<name>the return of Dr who</name>
<type>sc-fi</type>
</book>
<book>
<author>
<name>author2</name>
<gender>f</gender>
</author>
<name>something2</name>
<type>sc-fi</type>
</book>
...

say i was looking for all the type of books written by aurthor2, using the following call the processor will skip the type which it sees before (e.g. sc-fi will not be shown as it already appears previously for author1)

<xsl:apply-template select="book[author='author2']][generate-id() = generate-id(key('books', type))]"/>

therefore i was wondering is there a way to overcome this? (as you suggested) i though by filtering the data first before using the mucheun method (via multiple predicate) could solve the problem, however, it seems it doesn't like. please could you give me some advise over this?

furthermore, i was wondering it is possible to use preceding-silbing axis whilst it requires to match more than one element? for example using keys you can do this:
<xsl:key name="books" match="book" use="concat(author/name, '::', author/gender)"/>
and grouping via <xsl:apply-template select="book[generate-id() = generate-id(key('books', concat(author/name, '::', author/gender)))]"/>

however using preceding axis algorithum

<xsl:for-each select="/book[not(concat(author/name, '::', author/gender)=preceding-sibling::book/???
i am lost at that point....

please help


Many thanks

Regards Kit