Issue 6454: 14.3: StateInvariant and ExecutionOccurrence (uml2-superstructure-ftf) Source: (, ) Nature: Uncategorized Issue Severity: Summary: 14.3: StateInvariant and ExecutionOccurrence are both subclasses of InteractionFragment. "Each interaction fragment is conceptually like an interaction by itself." [14.3.9] And, indeed, "An ExecutionOccurrence is an instantiation of a unit of behavior..." [14.3.4] But, "A StateInvariant is a constraint on ... state..." [14.3.17] That's not like an interaction by itself, nor like any interaction at all. We've mixed models of behavior with specifications of constraints on state. This is an example of a recurrent problem in the specification: subclasses that are not like their superclasses. ... Suggested resolution: Review the specification with this in mind and correct all improper subtyping Resolution: Revised Text: Actions taken: November 7, 2003: received issue March 9, 2005: closed issue Discussion: It seems that the submitter missed the fact that the interactions metamodel is actually a syntactically-oriented metamodel rather than a semantically-oriented one. That is, the concepts in this metamodel are derived from their role in the abstract syntax of interactions rather than the semantic entities that they represent. Thus, the concept of InteractionFragment is a syntactical notion (something that is attached to a lifeline), and not a semantic notion. So, in this case, there is no problem of subclasses not being like their superclasses, since each specialization of InteractionFragment inherits all the things from its superclass. Perhaps the interactions metamodel could have been organized differently, but that kind of rearrangement is out of scope for an FTF. Disposition: Closed, no change End of Annotations:===== eply-To: Joaquin Miller X-Sender: joaquin%joaquin.net@secure.cnchost.com X-Mailer: QUALCOMM Windows Eudora Version 5.1 Date: Fri, 07 Nov 2003 10:14:50 -0800 To: issue@omg.org, issues@omg.org From: Joaquin Miller Subject: UML 2 Superstructure issues [Juergen, this may be a duplicate, if both e-mail addresses work. ] [Since you are sending issues in batches, i guess it is ok to submit them in batches. Here goes.] UML 2 Superstructure issues Joaquin Miller X-Change Technologies joaquin.no.spam@acm.org --------------------------------------------- 14.3: StateInvariant and ExecutionOccurrence are both subclasses of InteractionFragment. "Each interaction fragment is conceptually like an interaction by itself." [14.3.9] And, indeed, "An ExecutionOccurrence is an instantiation of a unit of behavior..." [14.3.4] But, "A StateInvariant is a constraint on ... state..." [14.3.17] That's not like an interaction by itself, nor like any interaction at all. We've mixed models of behavior with specifications of constraints on state. To: uml2-superstructure-ftf@omg.org Subject: Re: Fw: Proposed resolutions for ballot 13 from Bran X-Mailer: Lotus Notes Release 6.0.2CF1 June 9, 2003 From: Branislav Selic Date: Tue, 20 Apr 2004 09:51:37 -0400 X-MIMETrack: Serialize by Router on D25ML05/25/M/IBM(Release 6.0.2CF1|June 9, 2003) at 04/20/2004 09:51:39, Serialize complete at 04/20/2004 09:51:39 I am surprised that anyone would be offended by your claims, Joaquin, although it may be a matter of opinion to a degree. To me, something like InteractionFragment, which combines things as diverse as states, execution occurrences, and references to other interactions into a single general concept, is clearly primarily a syntactical notion not a semantic one. However, the dividing line between syntax and semantics is not alwyas an easy one to draw. For instance, is the notion of a state entry or exit point a syntactical or semantic one? To what extent is this made necessary by the specific formalism chosen versus the inherent properties of the phenomena that are being modeled? I think that this is a deep question and not an easy one to answer. However, I do think that most of the rest of the spec is semantically based rather than syntactically based, but I may think differently if I spent more time thinking about it. Bran Branislav Selic/Ottawa/IBM@IBMCA 04/20/2004 08:07 AM To uml2-superstructure-ftf@omg.org cc Subject Fw: Proposed resolutions for ballot 13 from Bran Forwarded from Joaquin...Bran ----- Forwarded by Branislav Selic/Ottawa/IBM on 04/20/2004 08:06 AM ----- Joaquin Miller 04/20/2004 12:16 AM Please respond to Joaquin Miller To Branislav Selic/Ottawa/IBM@IBMCA cc Juergen Boldt Subject Fwd: Re: Proposed resolutions for ballot 13 from Bran Bran, kindly forward this for me. Thanks. To: uml2-superstructure-ftf@omg.org, mu2i-ftf@omg.org, ocl2-ftf@omg.org From: Joaquin Miller Date: Mon, 19 Apr 2004 16:42:10 -0700 Subject: Re: Proposed resolutions for ballot 13 from Bran Bran wrote: ... please review the proposed resolution carefully and tell me what you think or suggest an alternative resolution if you are unhappy with mine. 6454 (Joaquin) The resolution tells us: "It seems that the submitter (me) missed the fact that the interactions metamodel is actually a syntactically-oriented metamodel rather than a semantically-oriented one." I like that (and the elaboration of the fact included in the resolution). I have more than once heard that vehemently denied when i have called attention to it in ADTF meetings. Some folks even appeared to be offended. I'm content to have the record show that this is, in fact, a fact. It's not only good to be straight about this; it also makes life much easier for readers of the model. I'd be even happier if text to this effect were inserted at an appropriate place in the specification. (At the front, if this is true of the entire metamodel, or at the beginning of interactions, if we mix and mismatch. (Clearly, such a text is even more valuable, if we do mix.)) 6459 (Joaquin) OK by me. To: uml2-superstructure-ftf@omg.org Subject: Fw: ,gi, ,, Issue 6454 model of what? X-Mailer: Lotus Notes Release 6.0.2CF1 June 9, 2003 From: Branislav Selic Date: Tue, 20 Apr 2004 14:09:58 -0400 X-MIMETrack: Serialize by Router on D25ML05/25/M/IBM(Release 6.0.2CF1|June 9, 2003) at 04/20/2004 14:10:02, Serialize complete at 04/20/2004 14:10:02 From Joaquin... ----- Forwarded by Branislav Selic/Ottawa/IBM on 04/20/2004 02:09 PM ----- Joaquin Miller 04/20/2004 01:48 PM Please respond to Joaquin Miller To Branislav Selic/Ottawa/IBM@IBMCA cc uml2-superstructure-ftf@omg.org Subject ,gi, ,, Issue 6454 model of what? Thanks, Bran. [Kindly forward to the list, unless you get a copy from the list.] Bran wrote: However, I do think that most of the rest of the spec is semantically based rather than syntactically based, but I may think differently if I spent more time thinking about it. This reader can form the impression that the rest of the spec wants to be read just as you say. So i feel the resolution of this issue should be to add text at the beginning of Interactions explaining that this metamodel is different from the rest, and how it differs. It's not only good to be straight about this; it also makes life much easier for readers of the model. >From the draft for a resolution: This is an example of a recurrent problem in the specification: subclasses that are not like their superclasses.