Re: New syntax spec

>>>Rick Jelliffe said:
> But surely that is wrong, because
>    [attributes]=set(idAboutAttr?, bagIdAttr?, propertyAttr*)
> then expands to
>    [attributes]=set(idAboutAttr?, bagIdAttr?, (typeAttr | propAttr)*)
> which, allowing multiple typeAttr, is impossible?
> An XML element can only have one attribute with a single name***
> Do you mean this:
>    [attributes]=set(idAboutAttr?, bagIdAttr?, typeAttr?, propAttr* )

True (and this is a bit better) but since propAttr also matches
"rdf:type", how do I express that when rdf:type is seen, typeAttr
should be accepted in priority to propAttr.

Maybe I could define propAttr as all namespaced attributes except for
those with namespace 'rdf-ns' and local name 'type'.  Which is rather
a mess. 

> Also, I think by "any" namespace you always mean "any except RDF".
> And does this syntax exhaust the RDF namespace? 

No I don't mean except RDF.  You can put RDF-namespaced properties
anywhere that properties can be used - attributes or elements.


> *** Actually, one could think of getting around this using
>   <x  xmlns:r1="namespace for rdf" xmlns:r2="namespace for rdf"
>         r1:type=" ..." r2:type="..." />
> but Namespace in XML explicitly disallows this in 5.3
> "Uniqueness of Attributes"

Yes, there would be different solutions if the standards upon which
we are based were not the way they are.  C'est la vie.


Received on Thursday, 13 September 2001 09:01:23 UTC