|
|
Re: Executable UML [message #1263128 is a reply to message #1262968] |
Tue, 04 March 2014 08:00 |
Dan George Messages: 9 Registered: December 2012 |
Junior Member |
|
|
Rafael, yes, a Property can have an isID property which is specified using property-modifier notation.
+idNumber : int {id}
This is close to EUML but it doesn't allow for multiple identifier sets.
The 2.5 beta spec mentions identifiers in the context of database keys or id attributes in XML. Leon Starr points out that in EUML models the {I}, {I2}, etc. marks are for the purpose of analysis and not implementation. He advises that modeling all identifying constraints helps to avoid creating buggy programs.
I could create an <<xuml>> stereotype and give it properties identifierSet and attributeOrigin. If identifierSet == "null" then non-identifier; otherwise, identifierSet value specifies the set membership. If attributeOrigin = ["native" | <relation-name>].
Does Papyrus provide a way for me to cause the stereotype values to be rendered as property-modifiers in the notation? E.g., +idNumber : int {I, R1}.
[Updated on: Tue, 04 March 2014 08:02] Report message to a moderator
|
|
|
|
|
|
|
Re: Executable UML [message #1360533 is a reply to message #1263540] |
Fri, 16 May 2014 21:30 |
|
Tags, or "tagged values" were a UML1 extension mechanism that allows for arbitrary text to be added to an element. There was not syntax associated, asides from being displayed in curly brackets.
The concept of tags has been replaced by stereotypes in UML2, which now have more semantics associated with them. For backward compatibility, some early UML2 tools simply provided the ability to enter anything in a special, no-syntax, stereotype property (along with provided formal definitions).
/Charles Rivet
|
|
|
Powered by
FUDForum. Page generated in 0.03963 seconds