On 11 Dec 2001, Christian Nentwich wrote:
>
> > Cool. Xalan is part of our lib/ directory, I think, so go ahead and
> > add your comparison test driver permanently, if you like.
> > > A bit of after-work hacking going on here.. hope I'll be able to commit
> > > all this bug-free tomorrow.
>
> I'm actually not going to commit these test classes because a) you have
> xalan 2 in your lib directory and I just converted some Xalan 1 classes
> and b) Xalan 2 makes variable assignment such a pain in the **** that I
> cannot be bothered.
>
> It's always struck me as a bit strange that the Xalan guys don't
> consider even the most basic use cases (variable assignment) and make
> them straightforward.
>
> One thing about the test cases: Your use of the context variables is
> rather strange. You evaluate the "context" attribute and then make every
> node the context node in turn.. which makes some tests impossible with
> my fix applied.
>
> For example,
>
> <context select="/PLAY/ACT/SCENE[1]">
> <valueOf select="count(descendant::*)+1">163</valueOf>
> </context>
>
> this test will never work, because /PLAY/ACT/SCENE[1] returns 5 nodes,
> and for each node the value of count(descendant::*) will be different.
> Hence it doesn't make sense to iterate through a context node set.
We just need to further specify with proximity predicates, I think.
/PLAY[1]/ACT[1]/SCENE[1]
or similar, to create a path to a single unique node.
-bob
_______________________________________________
Jaxen-interest mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jaxen-interest
