"Henry S. Thompson" wrote: > <snip> > > I've _just_ fixed this bug in the working draft. The intention has > always been (and XSV at least implements) that _unlike_ for content > models, if you omit an attribute altogether it _does_ get copied when > deriving by restriction. So to get rid of an attribute, you need to > use use="prohibited". Thanks. I'm curious, what will be the behaviour for the following definition <complexType> <attribute name="foo" use="prohibited"/> <anyAttribute/> </complexType> Will this satisfy constraints on schemas in the next working draft? If so, can I use foo in an instance? Ie does the prohibition just remove the attribute from my use pairs available, or does it specifically prevent its use in an instance, even by a wildcard? On a side note, is it mentioned in any of the rules that the below is ambiguous and not allowed? <complexType> <attribute name="foo"/> <anyAttribute/> </complexType> ct-props-correct.3 says I can't have identical attribute use pairs, but nothing about an attribute use pair being in the namespace of the wildcard. cvc-complex-type.1.3 says to match an attribute infoItem look in the pairs OR wildcard. What if it's in both? cheers, mick.Received on Monday, 19 February 2001 19:28:04 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 16 March 2009 11:13:16 GMT