> For one thing, the length facet of QName type is not defined in the spec.
> Did you mean something like this?
That's a good example.
> Also, what we may want to do?
>
> (1) prohibit QName type from having the length facet?
> (2) allow implementations to reject it?
> (3) encourage implementations to reject it?
I was thinking more of users than implementors. We would be telling users
of RELAX NG "It's better not to use a length facet on QName in RELAX NG
schemas, because it's not well-defined and so it probably won't
interoperate." This document isn't a spec with conformance requirements
like our other documents; it has more of an advisory role. As far as
implementors are concerned, I am not sure what our advice should be, other
than "Work to resolve this issue with the W3C".
>
> By the way, the length facet is also not defined for anyURI and NOTATION.
>
>
>
>
> Other things that may be of interest.
>
> - the language type and RFC1766 are inconsistent.
> http://lists.w3.org/Archives/Public/www-xml-schema-comments/2001JanMar/03
> 90.html
>
> - non-European cannot use the "time" type.
> http://lists.w3.org/Archives/Public/www-xml-schema-comments/2001JanMar/03
> 66.html
Scary.
The issue is whether it is worth our trying to come up with a list of
things (type/facet pairs) in XML Schema Part 2 that RELAX NG users should
avoid.
James
