Issue 15240: Owning of interaction fragments is ambiguous when InteractionOperands are present (uml2-rtf) Source: Microsoft (Mr. Steve Cook, stcook(at)microsoft.com) Nature: Uncategorized Issue Severity: Summary: It is not clear from chapter 14 how interaction fragments are supposed to be owned when there are InteractionOperands present. It seems to be the case, but is not stated, that everything diagrammatically inside the operand should be owned by the fragment. This would, I think, give rise to the following consequences: 1. The top and bottom of each fragment and operand must be on the same Lifeline or Execution. A fragment cannot span different executions or have its boundaries cover different levels of execution nesting. 2. Everything inside of a fragment/operand must be entirely contained by the fragment/operand. This includes both sides of a message, all nested fragments, interaction uses, and the top and bottom of execution specifications. However it appears to be a valid instance of the metamodel to parent arbitrary fragments at any level of nesting, which would enable these constraints to be violated. The specification should confirm these ownership constraints. This would best be done in conjunction with the sentence “InteractionOperand contains an ordered set of InteractionFragments” in section 14.3.16, which should state exactly which InteractionFragments must be owned by the InteractionOperand. Different vendors’ interpretations of this ambiguity can cause interoperability problems. Resolution: Revised Text: Actions taken: May 4, 2010: received issue Discussion: End of Annotations:===== m: Steve Cook To: "issues@omg.org" CC: "uml2-rtf@omg.org" , "juergen@omg.org" Subject: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Thread-Topic: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Thread-Index: Acrrg9ttItdpL01sTrytywehobbdjw== Date: Tue, 4 May 2010 12:18:10 +0000 Accept-Language: en-GB, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: It is not clear from chapter 14 how interaction fragments are supposed to be owned when there are InteractionOperands present. It seems to be the case, but is not stated, that everything diagrammatically inside the operand should be owned by the fragment. This would, I think, give rise to the following consequences: 1. The top and bottom of each fragment and operand must be on the same Lifeline or Execution. A fragment cannot span different executions or have its boundaries cover different levels of execution nesting. 2. Everything inside of a fragment/operand must be entirely contained by the fragment/operand. This includes both sides of a message, all nested fragments, interaction uses, and the top and bottom of execution specifications. However it appears to be a valid instance of the metamodel to parent arbitrary fragments at any level of nesting, which would enable these constraints to be violated. The specification should confirm these ownership constraints. This would best be done in conjunction with the sentence .InteractionOperand contains an ordered set of InteractionFragments. in section 14.3.16, which should state exactly which InteractionFragments must be owned by the InteractionOperand. Different vendors. interpretations of this ambiguity can cause interoperability problems. From: "Darren Kumasawa" To: "Steve Cook" , Cc: , Subject: Re: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Date: Thu, 6 May 2010 19:17:20 -0700 X-Mailer: Microsoft Outlook Express 6.00.2900.3598 X-Auth-Type: Internal IP X-Source-IP: rcsinet15.oracle.com [148.87.113.117] X-CT-RefId: str=0001.0A090205.4BE37874.017C:SCFMA4539811,ss=1,fgs=0 Hi Steve, I think that I understand your position, but could you please clarify how this relates to the example diagram in Figure 14.12 in section 14.3.3 for Combined Fragment in the 2.3 UML spec. For the ExecutionSpecification on lifeline ob1:C1 (the lifeline on the left) I think all of the MessageOccurrenceSpecifications inside this ExecutionSpecification will be owned by the 2 operands of the "alt" CombinedFragment. However, the "start" and "end" ExecutionOccurrenceSpecifications and the receiving MessageOccurrenceSpecification of the "opti" message will be owned by the Interaction since they are above and below the CombinedFragment. So even though parts of the ExecutionSpecification are owned by the CombinedFragment, the entire ExecutionSpecification from top to bottom is not owned by the CombinedFragment. Does my interpretation match with your proposal? Thanks, Darren Kumasawa Oracle Corporation ----- Original Message ----- From: Steve Cook To: issues@omg.org Cc: uml2-rtf@omg.org ; juergen@omg.org Sent: Tuesday, May 04, 2010 5:18 AM Subject: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present It is not clear from chapter 14 how interaction fragments are supposed to be owned when there are InteractionOperands present. It seems to be the case, but is not stated, that everything diagrammatically inside the operand should be owned by the fragment. This would, I think, give rise to the following consequences: 1. The top and bottom of each fragment and operand must be on the same Lifeline or Execution. A fragment cannot span different executions or have its boundaries cover different levels of execution nesting. 2. Everything inside of a fragment/operand must be entirely contained by the fragment/operand. This includes both sides of a message, all nested fragments, interaction uses, and the top and bottom of execution specifications. However it appears to be a valid instance of the metamodel to parent arbitrary fragments at any level of nesting, which would enable these constraints to be violated. The specification should confirm these ownership constraints. This would best be done in conjunction with the sentence .InteractionOperand contains an ordered set of InteractionFragments. in section 14.3.16, which should state exactly which InteractionFragments must be owned by the InteractionOperand. Different vendors. interpretations of this ambiguity can cause interoperability problems. From: Steve Cook To: Darren Kumasawa , "issues@omg.org" CC: "uml2-rtf@omg.org" , "juergen@omg.org" Subject: RE: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Thread-Topic: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Thread-Index: Acrrg9ttItdpL01sTrytywehobbdjwCB+8MiAA+pDyA= Date: Fri, 7 May 2010 09:53:32 +0000 Accept-Language: en-GB, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: >parts of the ExecutionSpecification are owned by the CombinedFragment, the entire ExecutionSpecification from top to bottom is not owned by the CombinedFragment. I wouldn.t say it like that. I don.t think of OccurrenceSpecifications that are between the start and end of an ExecutionSpecification as .parts of the ExecutionSpecification.; the ExecutionSpecification is a single object that is itself owned by the Interaction. >Does my interpretation match with your proposal? With the clarification above, yes. -- Steve From: Darren Kumasawa [mailto:darren.kumasawa@oracle.com] Sent: 07 May 2010 03:17 To: Steve Cook; issues@omg.org Cc: uml2-rtf@omg.org; juergen@omg.org Subject: Re: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Hi Steve, I think that I understand your position, but could you please clarify how this relates to the example diagram in Figure 14.12 in section 14.3.3 for Combined Fragment in the 2.3 UML spec. For the ExecutionSpecification on lifeline ob1:C1 (the lifeline on the left) I think all of the MessageOccurrenceSpecifications inside this ExecutionSpecification will be owned by the 2 operands of the "alt" CombinedFragment. However, the "start" and "end" ExecutionOccurrenceSpecifications and the receiving MessageOccurrenceSpecification of the "opti" message will be owned by the Interaction since they are above and below the CombinedFragment. So even though parts of the ExecutionSpecification are owned by the CombinedFragment, the entire ExecutionSpecification from top to bottom is not owned by the CombinedFragment. Does my interpretation match with your proposal? Thanks, Darren Kumasawa Oracle Corporation ----- Original Message ----- From: Steve Cook To: issues@omg.org Cc: uml2-rtf@omg.org ; juergen@omg.org Sent: Tuesday, May 04, 2010 5:18 AM Subject: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present It is not clear from chapter 14 how interaction fragments are supposed to be owned when there are InteractionOperands present. It seems to be the case, but is not stated, that everything diagrammatically inside the operand should be owned by the fragment. This would, I think, give rise to the following consequences: 1. The top and bottom of each fragment and operand must be on the same Lifeline or Execution. A fragment cannot span different executions or have its boundaries cover different levels of execution nesting. 2. Everything inside of a fragment/operand must be entirely contained by the fragment/operand. This includes both sides of a message, all nested fragments, interaction uses, and the top and bottom of execution specifications. However it appears to be a valid instance of the metamodel to parent arbitrary fragments at any level of nesting, which would enable these constraints to be violated. The specification should confirm these ownership constraints. This would best be done in conjunction with the sentence .InteractionOperand contains an ordered set of InteractionFragments. in section 14.3.16, which should state exactly which InteractionFragments must be owned by the InteractionOperand. Different vendors. interpretations of this ambiguity can cause interoperability problems. From: "Darren Kumasawa" To: "Steve Cook" , Cc: , Subject: Re: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Date: Fri, 7 May 2010 11:54:02 -0700 X-Mailer: Microsoft Outlook Express 6.00.2900.3598 X-Auth-Type: Internal IP X-Source-IP: acsinet15.oracle.com [141.146.126.227] X-CT-RefId: str=0001.0A090208.4BE46211.00B5:SCFMA922111,ss=1,fgs=0 Hi Steve, Sorry, my last email didn't do a good job of distinguishing between diagram shapes and model elements. In the example diagram in Figure 14.12 in section 14.3.3 for Combined Fragment in the 2.3 UML spec: For the execution rectangle on lifeline ob1:C1 (the lifeline on the left) I think all of the MessageOccurrenceSpecifications inside combined fragment rectangle will be owned by the 2 operands of the "alt" CombinedFragment. However, the "start" and "end" ExecutionOccurrenceSpecifications and the receiving MessageOccurrenceSpecification of the "opti" message will be owned by the Interaction since they are above and below the CombinedFragment rectangle. So even though parts of the execution rectangle are owned by the CombinedFragment, the entire execution rectangle from top to bottom is not owned by the CombinedFragment. Thanks, Darren Kumasawa Oracle Corporation ----- Original Message ----- From: Steve Cook To: Darren Kumasawa ; issues@omg.org Cc: uml2-rtf@omg.org ; juergen@omg.org Sent: Friday, May 07, 2010 2:53 AM Subject: RE: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present >parts of the ExecutionSpecification are owned by the CombinedFragment, the entire ExecutionSpecification from top to bottom is not owned by the CombinedFragment. I wouldn.t say it like that. I don.t think of OccurrenceSpecifications that are between the start and end of an ExecutionSpecification as .parts of the ExecutionSpecification.; the ExecutionSpecification is a single object that is itself owned by the Interaction. >Does my interpretation match with your proposal? With the clarification above, yes. -- Steve From: Darren Kumasawa [mailto:darren.kumasawa@oracle.com] Sent: 07 May 2010 03:17 To: Steve Cook; issues@omg.org Cc: uml2-rtf@omg.org; juergen@omg.org Subject: Re: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present Hi Steve, I think that I understand your position, but could you please clarify how this relates to the example diagram in Figure 14.12 in section 14.3.3 for Combined Fragment in the 2.3 UML spec. For the ExecutionSpecification on lifeline ob1:C1 (the lifeline on the left) I think all of the MessageOccurrenceSpecifications inside this ExecutionSpecification will be owned by the 2 operands of the "alt" CombinedFragment. However, the "start" and "end" ExecutionOccurrenceSpecifications and the receiving MessageOccurrenceSpecification of the "opti" message will be owned by the Interaction since they are above and below the CombinedFragment. So even though parts of the ExecutionSpecification are owned by the CombinedFragment, the entire ExecutionSpecification from top to bottom is not owned by the CombinedFragment. Does my interpretation match with your proposal? Thanks, Darren Kumasawa Oracle Corporation ----- Original Message ----- From: Steve Cook To: issues@omg.org Cc: uml2-rtf@omg.org ; juergen@omg.org Sent: Tuesday, May 04, 2010 5:18 AM Subject: UML 2 Issue: Owning of interaction fragments is ambiguous when InteractionOperands are present It is not clear from chapter 14 how interaction fragments are supposed to be owned when there are InteractionOperands present. It seems to be the case, but is not stated, that everything diagrammatically inside the operand should be owned by the fragment. This would, I think, give rise to the following consequences: 1. The top and bottom of each fragment and operand must be on the same Lifeline or Execution. A fragment cannot span different executions or have its boundaries cover different levels of execution nesting. 2. Everything inside of a fragment/operand must be entirely contained by the fragment/operand. This includes both sides of a message, all nested fragments, interaction uses, and the top and bottom of execution specifications. However it appears to be a valid instance of the metamodel to parent arbitrary fragments at any level of nesting, which would enable these constraints to be violated. The specification should confirm these ownership constraints. This would best be done in conjunction with the sentence .InteractionOperand contains an ordered set of InteractionFragments. in section 14.3.16, which should state exactly which InteractionFragments must be owned by the InteractionOperand. Different vendors. interpretations of this ambiguity can cause interoperability problems.