The main use of a void type is that it helps you construct a system with closure properties, so you can have operators on types (e.g. the intersection of two types) where the result of the operation is always a legitimate type.
Use-cases are the bane of orthogonality. Any system with good orthogonality can do things for which it is very difficult to find a use case. What’s the use case for adding two negative numbers? Design focused excessively on use-cases leads to a lack of orthogonality.
XSD 1.1 introduced xs:error which is a void simple type. It’s defined similarly to <xs:choice/> as a union with no member types. The specific use case for introducing it was for conditional type assignment, it allows you to have the last alternative in a list of alternatives say “if none of the above conditions is true, then the type is xs:error, meaning no content is accepted”.
Michael Kay Saxonica
Hi Folks, Thanks for the outstanding responses. I have learned that an empty choice:
<xs:complexType name="ValidationKiller">
<xs:choice/>
</xs:complexType> has no valid content, i.e., the set of valid content is the empty set, {}. When would one use such a complex type? Can you think of a use-case for it? /Roger
|
