Re: The identity transform and attributes

> I've been caught out before doing things like:
I think that's a different problem though, unrelated to the fact that
attributes don't take children.


> ...which doesn't work - you have to use name()
at least you don't have to use name in xslt2 (although you still can't
use self::)
"@* except @foo"
for example works and probably is more readable than the self:: test anyway.

> Again, I don't know why the self axis isn't available for attribute
> nodes... any ideas?

The confusing thing is that attribute nodes are on the self axis, if you
are on an attribute then self::attribute() selects the current node,
it's "just" that the QName node test always goes via the "principle node
type" kludge/feature and they decided to keep that being fixed for each
axis to a single node type, element in the case of self::. personally I
think it would have been much simpler if axis::foo had selected any node
with name foo that is in the axis axis:: It's hard to make any examples
where that formulation would have been different in practice from what
was specified, except for making self::foo select an attribute named foo
if it is the current node, which would have been an improvement.
in xpath2 of course you can go self::attribute(foo) but I can't see what
harm would have resulted from allowing (in xslt 1) self::foo for this.
In XPath 2 I think, given that attribute(foo) and the execpt operator
make the use for this a lot more rare anyway, they were right to keep
things compatible (but strange)


David

________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.

This e-mail has been scanned for all viruses by Star. The service is
powered by MessageLabs. 
________________________________________________________________________