CORBASequence uses multiplicity at the member members to define the sequence upper bound.
The most frequent case of CORBAArray is the single dimensional array.
Proposal: Permit using the members multiplicity for this common case in a similar way as for bounded sequence.
When using this alternative way of modeling:

• The multiplicity lower bound and upper bound shall both be set to the same value - the array size.
• The stereotype tag index shall not be used; in other words, usage of attribute multiplicity shall not be mixed with the index tag.

Modeling of multi dimensional arrays is not in affected by this proposal.

According to the IDL 4.2 grammar, the bound number of bounded sequence / bounded string and fixed array size is a
positive_int_const which resolves to const_expr.
In the examples for bounded sequence an array, the specification only shows literal constants used for the sizes.
It is not defined how to use expressions, such as a reference to an IDL const declaration, as the size.
At the least, it should be required to model a UML Dependency association from the CORBASequence / CORBAArray type to the const declaration.

The specification omits support for the IDL native declaration.

Issue: The VMM elements derived from CORBAWrapper need not be stereotypes of
Class; Classifier would suffice. Cleaning this up would require a small
rearrangement of the VMM. This only affects abstract stereotypes, so it
doesn't change how one actually models.

Recommendation: Change the VMM accordingly.

resolved

The name of the model element for CORBAUnions may conflict with
legitimate member names of the union.

Recommendation: Instead of specifying a naming convention for the model
element representing a switch, define a stereotype for a switch and use this
stereotype to denote that a model element represents a switch. There will
need to be one stereotype that extends AssociationEnd and one that extends
Attribute, because the model element representing a switch is an
AssociationEnd if the switch's type is non-primitive, and is an Attribute if
the switch's type is primitive.

resolved

Issue: The submission's coverage of CORBA primitive types (section 6.5.1.1)
does not cover the long double primitive type.

Recommendation: Add long double to the coverage of CORBA primitive types.

long double will be added to the set of primitives in the same style as long long.

Recommedation: If possible, synchronize the FTF report with the UML RTF
report.

I have no way to resolve this issue, and I suggest we close this issue without further action.

Issue: Namespace rules have been changed in the latest CORBA core
specification. Specifically, a parent and child can't have the same name.
This restriction holds only for children and does not hold for grandchildren
and more distant descendants.

Recommendation: Fix the formal constraints about namespace containment
specified in the submission.

An English constraint and equivalent OCL expression will be inserted to reflect this IDL constraint.

Recommendation: Specify this (probably CORBA 2.4) and flag IDL features that
the submission covers but that haven't been made available yet. (Since
local has been placed in 2.4, attributes raising exceptions is the only
feature that falls into this category). The conformance language for
forward engineering tools should be correspondingly tightened.

No Data Available

in UML Profile for CORBA, where it describes the UML notation examples for mapping of CORBA sequences/arrays...there is some significant confusion on my part with respect to the notation examples given. See Figure 3-27 for an example. In this figure the class stereotyped as <<CORBAAnonymousSequence>> has a composition association with a class stereotyped as <<CORBAPrimitive>>. The latter class has the following text appearing in the name compartment: "CORBA::short". I am bewildered at how the type can be represented in the name compartment of the class. I am trying to use a tool to auto-generate IDL from UML, and have no way to represent this, and I am unsure if this is even valid as per the UML specification. Is this legal, or is necessary to have an attribute that is of type CORBA::short in the attribute compartment?