Re: [xsl] Question about grouping

[Michael Kay <mike@xxxxxxxxxxxx>]

> While Hugh's answer might work, the method used in his first for-each select is kind of an odd way of doing Muenchian grouping in XSLT 1.0. Here's the approach with the standard method:...

Both methods

count(A) = count(A|B)

and

generate-id(A) = generate-id(B)

are perfectly acceptable and "standard" ways of comparing node identity 
in XSLT 1.0. Years ago Jeni did some measurements and found that in some 
processors the first was faster, in others the second was faster. Both 
are in common use.

(I seem to remember that the generate-id() method was much slower in 
Saxon, but I then optimized it in the next release to compare node 
identity directly without actually computing the IDs.)

Michael Kay
Saxonica