>>>>> "Martin" == Martin Honnen <mahotrash@y...> writes:

    Martin> Michael Powe wrote:

    >> I have input structure like this:

    >> <entry> <dimension name="dimName" value="dimValue"/> <dimension
    >> name="dimName2" value="dimValue2"/> <metric name="metName"
    >> value="metValue"/> <metric name="metName2" value="metValue2"/>
    >> </entry>
    >> 
    >> I need to transform this entry into this format:
    >> 
    >> <row> <field name="dimName" value="dimValue"/> <field
    >> name="metName" value="metValue"/> <field name="metName2"
    >> value="metValue2"/> </row> <row> <field name="dimName2"
    >> value="dimValue2"/> <field name="metName" value="metValue"/>
    >> <field name="metName2" value="metValue2"/> </row>
    >> 
    >> This seems to be a nontrivial problem.  I can easily loop
    >> through the entry node and retrieve the dimension/metric
    >> attributes and create the field elements.  However, I don't
    >> understand how in XSLT to process the dimension, jump to the
    >> metrics and process those, then return to the next dimension
    >> element, and so forth.  Further, the 2 + 2 in this case is
    >> simply a test document.  The production data could have 1-7
    >> dimension elements and 1-50 metric elements in each entry node.

    Martin> I would simply process each dimension element and
    Martin> transform it into a row and then inside each row you
    Martin> process the element itself and all its metric siblings:

    [ ... ]

    Martin> As the dimension element is processed twice you need to
    Martin> use a mode to distinguish those steps but other than that
    Martin> there is nothing complicated about that stylesheet.

Of course, the complication lies in the ability to correctly decompose
the task!  ;-)

Thank you very much.  As always, the key to a good solution is to have
properly thought through the problem.

And, you've given me much to think about.

Thanks.

mp
-- 
cat: $HOME/.signature: No such file or directory