MOF should publish a convenience document that pre-merges all the different packages
The simplification and alignment between UML and MOF in the 2.4 series is incomplete.
In particular, the extensions that MOF adds to UML are missing in UML.
The MOF extensions that are missing in UML mean that statements like the one below in UML 2.5, section 6.2 are technically incorrect:
Since version 2.4.1 a MOF 2.x metamodel, including the UML 2.x metamodel, is a valid UML 2.x model. This was a substantial simplification and alignment compared to earlier versions. It is expected that future versions of MOF and UML will continue to be aligned in this manner.
For example, UML has no mechanism to specify the information about MOF::Extension::Tag. Without this information, it is currently not possible to fully rely on the above statement to use UML as a language for representing models of UML itself or parts of it such as the PrimitiveTypes library.
One fairly simple option would be to define a MOF profile with stereotypes corresponding to the contents of the MOF-specific extensions of UML
The uml diagram in chapter 10.4 MOF:Common specifie that ReflexiveCollection remove method return a Boolean. Just after, when describing this method (chapter 10.5) it is set to Object.
UML uses operation redefinition quite extensively, but MOF says nothing about what this means, and UML itself leaves it rather open see for example issues 17924 and 15499. UML 2.5 beta leaves it even more open than 2.4.1, which means that MOF needs to be specific for UML to be well-defined.
My suggestion is that MOF requires parameters in redefined operations to have the same number, type, multiplicity, uniqueness and ordering as the parameters in the operation being redefined.
The portion of text ""An Extent is a context in which an Element in a set of Elements in a set can be identified"".
I think that this way is better: ""An Extent is a context in which an Element in a set of Elements can be identified""
That way, I can understanding. To more detail, in my language (portuguese), the Google shows a concise translation.
Regarding to spec:
§13.4 Object (from CMOF Reflection)
"CMOF Reflection adds the following extra operations.
invoke(op:Operation, arguments : Argument[0..*]) : Object[0..*]"
But this operation is present into http://www.omg.org/spec/MOF/20131001/MOF.xmi file for MOF::Reflection::Object :
<packagedElement xmi:type="uml:Class" name="Object" xmi:id="_MOF-Reflection-Object">
...
<ownedOperation xmi:type="uml:Operation" name="invoke" visibility="public" xmi:id="_MOF-Reflection-Object-invoke">
...
</ownedOperation>
Moreover it uses <type xmi:idref="_MOF-CMOFReflection-Argument"/> from CMOF.
Thanks
The text "An Extent is a context in which an Element in a set of Elements in a set can be identified." contains sentence fragment duplication.
In MOF 2.4, CMOF reflection allows you to create a link as an instance of a single association and to test two links for equality. However, MOF 2.4 seems to be silent on the semantics of association generalization.
According to the UML spec, if I have two associations A1 and A2, where A2 specializes A1, and I create a link between two objects o1 and o2 as an instance of A2, a link of A2 is also a link of A1. From 7.3.20 Generalization: “Each instance of the specific classifier is also an indirect instance of the general classifier.”
However, CMOF semantics defines link equality so it requires association equality as well. This appears to conflict with UML semantics.
Does this matter? Well, the UML metamodel does have association generalizations in it, where needed to make property redefinition syntactically valid. For such association generalizations, if we accept the MOF semantics, then they must be purely a syntactic convenience, i.e. they do not imply the link classification that the UML spec says that they imply.
But there is a problem. Let’s look more closely at association specialization in the UML metamodel. In Fig 13.13 of UML 2.4 there is a property Interval::min that is redefined as TimeInterval::min. The opposites of these properties are also redefined and are association-owned, so the associations specialize in order to maintain well-formedness. I can confirm this is true in the UML metamodel. If I set TimeInterval::min to a particular value, then I am instantiating A_min_timeInterval. Because of the MOF semantics, I am not instantiating A_min_interval. That gives anomalies. If I were given an instance x that I knew to be at least an Interval, and used CMOF reflection linkedElements(A_min_interval, x, true) to find the linked elements, I will not find its min value in the case where x is actually a TimeInterval. I count that as a bug.
In summary, the CMOF API that allows links to be explicitly manipulated and navigated should be defined so that a link of a sub-association is also a link of its super-associations.
Clause 2 specifies a requirement that compliant products shall support certain technology mappings, but there is no reference either here or in Clause 3, Normative references, to sources of specifications of those mappings.
There is an inconsistency in the current spec between link equality and link delete. Link equality only checks for the end values and the association, whereas link delete says: “This may leave the same elements associated by other links for this Association”, implying more than one distinguishable link per pair of elements. This should be resolved according to the fact that in OCL and UML collections resulting from navigating associations can be bags, i.e. contain the same element more than once. Links must be distinguishable individually, not simply by equality of their ends. This will imply that links have some identity criteria: uuids would seem to fit the bill perfectly.
There is no reflective access to uuids, and no mechanism is defined to assign uuids to objects
Section 9.2 contains a set of constraints these should be rationalized with the ‘main” constraints in 12.4 (CMOF) and 14.3 (CMOF). A lot of them are in fact redundant (e.g. UML constraints) or not needed
URIExtent should provide the capability of accessing links as well as elements by URI. This will enable SMOF multiply-classified links to be serialized with their uuids.
The file http://www.omg.org/spec/MOF/20131001/MOF.xmi starts with:
<xmi:XMI xmlns:mofext="http://www.omg.org/spec/MOF/20131001" xmlns:uml="http://www.omg.org/spec/UML/20131001" xmlns:xmi="http://www.omg.org/spec/XMI/20131001">
<packagedElement xmi:type="uml:Package" xmi:id="_MOF" name="MOF">
...
Is it correct to have <packagedElement> as root Model element ?
Why isn't it :
<xmi:XMI xmlns:mofext="http://www.omg.org/spec/MOF/20131001" xmlns:uml="http://www.omg.org/spec/UML/20131001" xmlns:xmi="http://www.omg.org/spec/XMI/20131001">
<uml:Package xmi:id="_MOF" name="MOF">
...
Thanks
"remove(object: Object): Object
Removes the specified object from the collection. Returns true if the object was removed."
Maybe the operation should return a Boolean.
Agreed. Correct the operation signature to return Boolean
Who owns LinkSlots? When an association end is owned by a Classifier, are there two slots for its instances (one for the link and one for the element) or only one?
In the abstract semantics there is a concept called LinkSlot, which is shown as weakly aggregated (white diamond) by AssociationInstance. White diamond has not meaning in this context. Is it possible that a LinkSlot may be owned either by the link or by the adjacent instance, depending on “navigability”?
The following sentence appears to be key: “Where the feature is a navigable end, then the ClassInstance Slot is consistent with the Link slot.” The reference to "navigable" is surely incorrect. What does "consistent" mean?
LinkSlots are owned by the AssociationInstance. Update diagram 15-1 to show this
correctly.
Per UML2.4.1, a
In 12.4, constraint [8] currently reads:
An EMOF metamodel is restricted to use the following concrete metaclasses from UML’s Kernel:
Class
Comment
DataType
Enumeration
EnumerationLiteral
Generalization
InstanceValue
LiteralBoolean
LiteralInteger
LiteralNull
LiteralReal
LiteralString
LiteralUnlimitedNatural
Operation
Package
Parameter
PrimitiveType
Property
The list includes InstanceValue but incorrectly omits InstanceSpecification.
InstanceSpecification must be included in the list because an InstanceValue requires an InstanceSpecification;
see UML2.4.1, 7.3.23:
instance: InstanceSpecification [1]
The instance that is the specified value.
Since the list includes Class and a Class can have Property features, an InstanceSpecification that is the value of an InstanceValue in EMOF may have to specify values for the instantiated Class' Property features. Therefore, the list should also include UML2.4.1's Slot as well.
The list should be corrected as follows:
Class
Comment
DataType
Enumeration
EnumerationLiteral
Generalization
InstanceSpecification
InstanceValue
LiteralBoolean
LiteralInteger
LiteralNull
LiteralReal
LiteralString
LiteralUnlimitedNatural
Operation
Package
Parameter
PrimitiveType
Property
In 14.4, constraint [10] currently reads:
A CMOF metamodel is restricted to use the following concrete metaclasses from UML’s Kernel:
Association
Class
Comment
Constraint
DataType
ElementImport
Enumeration
EnumerationLiteral
Generalization
InstanceValue
LiteralBoolean
LiteralInteger
LiteralNull
LiteralReal
LiteralString
LiteralUnlimitedNatural
OpaqueExpression
Operation
Package
PackageImport
PackageMerge
Parameter
PrimitiveType
Property
The list includes InstanceValue but incorrectly omits InstanceSpecification.
InstanceSpecification must be included in the list because an InstanceValue requires an InstanceSpecification;
see UML2.4.1, 7.3.23:
instance: InstanceSpecification [1]
The instance that is the specified value.
Since the list includes Class and DataType, both of which can have Property features, an InstanceSpecification that is the value of an InstanceValue in CMOF may have to specify values for the instantiated Class' or DataType's Property features. Therefore, the list should also include UML2.4.1's Slot as well.
The list should be corrected as follows:
Association
Class
Comment
Constraint
DataType
ElementImport
Enumeration
EnumerationLiteral
Generalization
InstanceSpecification
InstanceValue
LiteralBoolean
LiteralInteger
LiteralNull
LiteralReal
LiteralString
LiteralUnlimitedNatural
OpaqueExpression
Operation
Package
PackageImport
PackageMerge
Parameter
PrimitiveType
Property
Slot
Accept the proposal
The document, which is about a mapping of MOF to XMI, does not define the initialisms MOF or XMI anywhere in the documents, either in Chapter 1 (Scope) or in a glossary, which does not exist in this document.
Chapter 1 starts with "XMI is a widely used interchange format..." and ends with "MOF is the foundation technology for describing metamodels..." but there is no mention of what the terms actually stand for, a significant omission in a formal document.
One may also question the use of "MOF 2 XMI" in the title. While "MOF2XMI" may be a common way to create a meta-abbreviation for a relationship between two concepts denoted by abbreviations, there is little justification to eschew the grammatically correct "MOF-to-XMI Mapping" (with hyphens or spaces) in the document title. Using the numeral "2" is actually ambiguous – does it mean "to" or does it refer to a version 2 of MOF?
Overall, this issue is more about casual editing than about correctness, but I believe it makes the specification weaker, editorially, than the usual OMG standard of documentation.
Make the change as suggested