W3C home > Mailing lists > Public > public-swbp-wg@w3.org > October 2005

Re: [SE] Suggestion of new note

From: Holger Knublauch <holger@SMI.Stanford.EDU>
Date: Mon, 10 Oct 2005 09:54:03 +0100
Message-ID: <434A2C2B.4050206@smi.stanford.edu>
To: public-swbp-wg@w3.org

>  Not sure what you are identifying with term "property" in the OO context.  

This may be my "fault", because I use the same term in the W3C draft. 
As far as I understand it, the term Property is used in the UML 2 spec 
to describe attributes or association ends.  These seem to correspond 
quite nicely to rdf:Properties.  I am not sure how widely used (and 
understood) the term property is among programmers though, and we may 
consider to reword this.  The alternative would be "attributes or 
association ends".


> This 
>  is because attributes and relations (uml:Association) are treated differently in OO.
>  The statement "Properties attached to single class" isn't quite right for OO attributes 
>  or relations.  Perhaps this would be better written "Attributes are defined locally
>  to a class".  Thus attributes with the same name in different classes do not 
>  denote the same attribute.

I have changed this in the document, following your suggestion.


>  Associations (relations) are defined with respect to particular classes at their 
>  AssociationEnds.  Thus Associations are attached to one or (usually) more classes.

Yes, but associations are rather syntactic sugar: what really counts in 
OO systems seems to be association ends, because the ends are mapped 
into variables/fields in the implementation language.  Many UML diagrams 
use associations only for visualization and don't even care to give them 
a name.  Also, there is no need to use associations to describe 
relations between classes - it is sufficient to use attributes (of class 
types).  This is frequently done when visual association arcs don't make 
sense (such as for associations across packages).

Holger
Received on Monday, 10 October 2005 08:54:21 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:09:44 UTC