Issue 10001: Merged Metam.:Property::class with redefinition of non-inherited property (uml2-rtf) Source: Adaptive (Mr. Pete Rivett, pete.rivett(at)adaptive.com) Nature: Uncategorized Issue Severity: Critical Summary: Merged metamodel has Property::class with redefinition of a non-inherited property ---------------------------------------------------------------------------------------------------------------------- In UML 2.1 we have the following: Kernel defines Class which inherits from Classifier, and has Class::ownedAttribute of type Kernel::Property. Composite Structures also defines Class which inherits from EncapsulatedClassifier which inherits from StructuredClassifier which inherits from Kernel::Classifier (curiously not Collaborations::Classifier in the same section). Now StructuredClassifier also defines property StructuredClassifier::ownedAttribute of type InternalStructures::Property So in the Merge, we have: L3::Class with property L3::Class::ownedAttribute of type L3::Property this will inherit from: L3::Classifier and L3::EncapsulatedClassifier, with the latter inheriting from L3::StructuredClassifier. And L3::StructuredClassifier will continue to have a property L3::StructuredClassifier::ownedAttribute. This would be inherited by L3::Class which has its own ownedAttribute. Hence there must be a redefinition L3::Class::ownedAttribute redefines L3::StructuredClassifier::ownedAttribute (there is). Likewise there must also be a generalization between the 2 associations (there is). However there is a change of the property ownership: at the subclass Property::class is owned by Property, and L3::A_ownedAttribute_structuredClassifier::structuredClassifier is owned by the Association. And there is no redefinition (or subsetting) between the two. Note that Figure 9.2 of ptc/06-04-02 does show a redefinition - but of "_structuredClassifier" with an underscore (not sure what that is supposed to mean). Proposed resolution: The Property::class property should be owned by the association (but still be navigable), and a redefinition needs to be added in section 9.3.12 {redefines structuredClassifier}. Add Property::classifier as a derived union and have all opposites of ?::ownedAttribute subset it. This way, access to a property's (owning) classifier can be obtained uniformly - note that a number of the OCL expressions are currently written (incorrectly) with this assumption. Resolution: Revised Text: Actions taken: July 26, 2006: received issue Discussion: End of Annotations:===== ubject: Urgent Issue on UML 2.1 Superstructure Date: Wed, 26 Jul 2006 09:50:52 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Urgent Issue on UML 2.1 Superstructure Thread-Index: Acaw06dcctxRGS0vRuyebdZiqMlztA== From: "Pete Rivett" To: , "Andrew Watson" Cc: "Branislav Selic" , "Kenneth Hussey" This has a major impact on our implementation so I would like to raise this as an Urgent Issue. Thanks to Kenn for the proposed resolution. Merged metamodel has Property::class with redefinition of a non-inherited property ---------------------------------------------------------------------------------------------------------------------- In UML 2.1 we have the following: Kernel defines Class which inherits from Classifier, and has Class::ownedAttribute of type Kernel::Property. Composite Structures also defines Class which inherits from EncapsulatedClassifier which inherits from StructuredClassifier which inherits from Kernel::Classifier (curiously not Collaborations::Classifier in the same section). Now StructuredClassifier also defines property StructuredClassifier::ownedAttribute of type InternalStructures::Property So in the Merge, we have: L3::Class with property L3::Class::ownedAttribute of type L3::Property this will inherit from: L3::Classifier and L3::EncapsulatedClassifier, with the latter inheriting from L3::StructuredClassifier. And L3::StructuredClassifier will continue to have a property L3::StructuredClassifier::ownedAttribute. This would be inherited by L3::Class which has its own ownedAttribute. Hence there must be a redefinition L3::Class::ownedAttribute redefines L3::StructuredClassifier::ownedAttribute (there is). Likewise there must also be a generalization between the 2 associations (there is). However there is a change of the property ownership: at the subclass Property::class is owned by Property, and L3::A_ownedAttribute_structuredClassifier::structuredClassifier is owned by the Association. And there is no redefinition (or subsetting) between the two. Note that Figure 9.2 of ptc/06-04-02 does show a redefinition - but of "_structuredClassifier" with an underscore (not sure what that is supposed to mean). Proposed resolution: The Property::class property should be owned by the association (but still be navigable), and a redefinition needs to be added in section 9.3.12 {redefines structuredClassifier}. Add Property::classifier as a derived union and have all opposites of ?::ownedAttribute subset it. This way, access to a property's (owning) classifier can be obtained uniformly - note that a number of the OCL expressions are currently written (incorrectly) with this assumption. Pete Rivett (mailto:pete.rivett@adaptive.com) CTO, Adaptive Inc. Hello House, 135 Somerford Road, Christchurch, BH23 3PY, UK Tel: +44 (0)1202 491243 Fax: +44 (0)1202 491241 http://www.adaptive.com ubject: typo in 10001 Date: Wed, 4 Oct 2006 09:27:02 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: typo in 10001 Thread-Index: Acbnxd1ZjJfC4y0vQpmnEb4XKD6iegAC6xzQ From: "Karl Frank" To: "Branislav Selic" , X-OriginalArrivalTime: 04 Oct 2006 16:27:04.0395 (UTC) FILETIME=[ED69C1B0:01C6E7D1] X-TM-AS-Product-Ver: SMEX-7.0.0.1499-3.6.1039-14730.003 X-TM-AS-Result: No--20.574500-8.000000-31 I see a mistake to fix in the wording of the resolution for 10001 "In figure 7.4, remove the constraint {..} from the association end on the composition assocciation running from Namespace to Packageable element" The association in question is not a composition. - Karl Frank. -------------------------------------------------------------------------------- From: Branislav Selic [mailto:bselic@ca.ibm.com] Sent: Wednesday, October 04, 2006 10:50 AM To: uml2-rtf@omg.org Subject: Draft ballot for URGENT fixes Importance: High Attached, please find a draft of the ballot with URGENT fixes to UML 2.1, which will result in UML 2.1.1. I remind you that this is an exceptional procedure that the OMG has set up to deal with absolutely critical implementation-blocking issues. All other open issues will be addressed in the context of the normal UML 2.2 RTF process. We will issue the official ballot most likely at noon EST tomorrow and follow that up with a 2-week voting period (using the usual RTF voting procedures). This URGENT ballot has four issues in it. Note that issue 10079 looks like a very large change but please note that it does not require any text in the spec to change, but only the XMI. The problem that it fixes has to do with certainly automatically generated association generalizations (and related association end redefinitions) in the metamodel that were inserted when the XMI was generated from the metamodel. These were put in for dealing with situations where one of the association ends was non-navigable (and, hence, owned by the association). It turns out that not only was there no need to generate those association generalizations but that this actually resulted in additional unexpected conflicts. By removing these generalizations and redefinitions, the conflicts disappear as does the problem raised by issue 10001 (although that one requires two minor changes to the spec). Issue 10074 simply removes the contradiction in the spec whereby certain model elements were specified as being owned by more than 1 composite (due to the fact that the multiplicity at the owning end was set to 1 instead of 0..1). So, again, there are no real semantic changes to the spec. Note that, in this version we do not include a resolution for issue 10376, since we are still waiting for the actual URIs to put in. However, you can see the nature of the resolution that will be provided in the text of the issue. So, please review this ballot and, if you have any concerns, raise them IMMEDIATELY with Pete and myself. As I said, the official ballot will likely be issued tomorrow. Finally, remember that to open the attached file, you will have to change the file suffix from .zap to .zip Regards, Bran Selic IBM Distinguished Engineer IBM Rational Software 770 Palladium Drive Kanata, Ontario, Canada K2V 1C8 ph.: (613) 591-7915 fax: (613) 599-3912 e-mail: bselic@ca.ibm.com