I've been re-reading the W3C Recommendation and although I still think
the definition for the case-order sorting given in it is misleading to say
the least, I have to recognise that if you keep reading the EXAMPLE given
in it clearly states that given A,B,a and b the sorting (Upper-case first)
would be A,a,B,b
And that's probably why all the implementators followed this rule.
According to the W3C Recommendation:
But implementers are not following that rule.
The rule given is that the case order affects the ording of characters
and that strings are ordered lexicographically based on that ordering.
This is _not_ what the implementations are doing.
Mike kindly gave the algorithm used in saxon 6. Given the length of time
its been used, I fear its to late to change it, but I can't see any
reading of the W3C rec that could justify such an algorithm.
Well actually there is one reading (as Mike pointed out) you could
assume that sorted lexicographically was being used as a colloquial
turn of phrase rather that specifying lexicographic ordering, but
I don't really see any justification for that (and it certainly never
occured to me before this thread that systems would do that)
David
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list