OCL21 — Provide specific notational support when testing stereotypes

Suggestion: Use the STAR character. Quotes could be used in case of blanks
characters in stereotype names.
Example: self.ownedElement.select(kindOf(Class) and *EJBEntity)
returns all the classes stereotyped by the EJBEntity stereotype or a derived
stereotype.
Example: self.ownedElement.select(kindOf(Class) and **EJBEntity)
returns all the classes stereotyped by the EJBEntity stereotype.

Stereotype support is certainly required, but I think much of the trouble is inadequate tooling. The examples can be realised by:
self.ownedElement->select(oclIsKindOf(Class)).oclAsType(Class) ->select(extension_EJBEntity <> null)
self.ownedElement->select(oclIsKindOf(Class)).oclAsType(Class) ->select(extension_EJBEntity.oclIsTypeOf(EJBEntity))
The clumsy ->select(oclIsKindOf(Class)).oclAsType(Class) 'idiom' is resolved by the selectByKind library operation from Issue 18829 to give
self.ownedElement->selectByKind(Class)->select(extension_EJBEntity <> null)
self.ownedElement->selectByKind(Class)->select(extension_EJBEntity.oclIsTypeOf(EJBEntity))
Given that UML specified the magic "extension_" and "base_" prefixes, it seems best to encourage rather than obscure them.
Disposition: Closed, no change