Re: Types defines in F & O and WXS 1.1

On Monday 23 January 2006 19:28, Jonathan Robie wrote:
> Michael Rys wrote:
> >The original discussion (and I thought agreement) was that WXS 1.1 would
> >simply refer to the XQuery types (including namespaces) where
> >appropriate.
> >
> >I have not looked at WXS 1.1 yet, but if that is not the case, then I
> >think there is a problem...
>
> The XML Schema Working Draft from 17 January [1] added anySimpleType,
> anyAtomicType, yearMonthDuration and dayTimeDuration, but added them
> in their own namespace, not in the XQuery namespace. That means that
> XML Schema and XQuery use different names for these types.
>
> I think this is a problem.

I think so too. This looks like an upcoming, potentially large problem. I 
wouldn't mind if this had serious consideration, working group discussion, 
but considering how this thread have development there's no need for me to 
worry about that.

> We can't say how we map from the Schema 1.1 
> types to our data model, because we do not yet do our mapping based on
> XML Schema 1.1. It also makes little sense to ask XML Schema to use
> our namespace for these types.
>
> I think the simplest thing to do is to have XQuery adopt the XML
> Schema 1.1 names for these types. That would also simplify naming of
> types in XQuery.

I agree, that's the solution to the problem I've been thinking of as well, 
although I have my doubts for whether it's possible. What I see can 
complicate is how to make F&O, XQuery, etc rely on something that will be in 
draft status when they reach recommendation status.

One solution is of course to wait until XML Schema goes recommendation. 
Another is perhaps to:

1) By close cooperation with the XML Schema WG ensure that the types in XML 
Schema 1.1 are compatible with the XQuery/XSL WGs' expectations when it is 
released.

2) Provide definitions of anyAtomicType and the two duration types in the 
XPath Data Model spec, in the XML Schema namespace(as opposed to the current, 
XDM namespace).

3) To explicitly mention the upcoming "double" specification of the types(in 
the XQuery specs, and the upcoming WXS 1.1), and handle it gracefully, by 
perhaps: 1) mention that the definitions are identical; and 2) to mention 
that the XQuery WG intend to release an errata which replaces XDM/F&O's 
definitions with references to WXS 1.1 once it is released. (Well, perhaps 
not exactly that, but hopefully the big picture is clear.)

Such a solution would obviously be far from optimal(if at all feasible), but 
the question is whether it's more worse than the upcoming double namespaces. 
No doubt that it is also an intrusive change at this stage, but that should 
be weighted against the future problems it avoids. Also, the type hierarchy 
becomes simpler.


		Frans

Received on Monday, 23 January 2006 21:02:09 UTC