Re: recalculating several models

> > This is basic XPath...
>
> Yes, but the example is using the id function to go outside of the
instance
> document itself. From my understanding, the <xf:instance> element is not
> considered part of the contained instance document, so XPathing to that
> level seems a bit odd. It almost seems as if we are saying that any
element
> in the entire XForms *document* may be referenced by id via the id()
> function.

That's exactly what I'm saying, but I may be wrong...
But why would xf:instance have an id attribute if you can't use it ? And you
can (almost) only use it with id().

> > this usage is suggested in chapter 4.2 of the XForms Draft.
>
> I see nothing in section 4.2 of the WD that leads me to believe that the
> id() function may be used in this fashion. It simply states that all XForm
> elements have an id which may be referenced from another context. I would
> not grow the assumption that using the id() function to cross-reference
> instance documents is what this section is speaking of.

Again, what would be the use of an id for xf:instance if you can't use the
id() function ?

IMHO, this was the solution Josef Dietl was talking about in his mail "RE:
evaluation context" from october 19th, 2001 (See
http://lists.w3.org/Archives/Public/www-forms/2001Oct/0041.html ).
Since this is pure speculation, I'd love to have feedback from Josef.

> I am not against using id() in this way but I disagree heavily at the idea
> that this usage is basic XPath or that the XPath specification speaks to
> using any of it's methods (id included) across multiple documents, as is
the
> case with multiple XForms instances.

Well, the problem is that there are two kinds of documents: the "real"
document (the file or what-ever) and the "XForms-point-of-view" documents
(the instances and so on). It might not be that basic ;-)

Maybe that another name should have been chosen for this function...

> Please advise,
>
> - Ryan

Anyway, that was not the point of this thread. Dependencies between
instances is a must IMHO, and I don't think the current draft deals with it.

I'd love to have feedback on this issue.

Regards,
Jérôme

Received on Monday, 7 January 2002 10:27:27 UTC