OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index]
RE: [ebxml-dev] Core components and UML

<Michel>We believe that UN/CEFACT and ebXML use as a baseline UML
(perhaps adapted to the specific purpose of EDI with the UN/CEFACT
version UMM). We have difficulty in understanding how to apply the UML
principles to the CCTS especially when we look at the core components.
All our analysts use extensively UML and one of the prinicpal interests
in ebXML and UN/CEFACT is the strong attachment to UML as it enables us
to remain business orientated rather than syntax orientated. The
standardisation of core components that can be used within UML seem to
be extremely rich to us and we therefore want to align our basic
constructs on their UN/CEFACT equivalents.
The basic principle of UML is that a class is the description of a set
of objects that share the same attributes, operations, methods,
relationships and semantics. When one looks at the principles of the
CCTS core components where a core component contains only the
information pieces necessary to describe a specific concept, one expects
a UML class and a core component to be very similar.
However, one does not understand the requirement within the CCTS that
obliges a basic core component only to be used in a aggregate core
component. Why can't an atomic part such as an attribut   (BCC)  be used
within any construct, as is possible in UML?</Michel>

A UML Class is equivalent to an ACC or an ABIE. A UML Class Attribute is
equivalent to a BCC or a BBIE. As far as I know UML Class Attributes are
always part of a Class and are not defined separately. Data Types are.

<Michel>Can one assume that an ASCC is merely an association between two
UML classes?</Michel>

Yes.

<Michel>There appears to be an inconsistancy between the use of a BCC
that can only be referenced through an ACC, however, in a message
construct one can mix and mingle both ABIEs and BBIEs as appears in
figure 4.2. Thus, since a BBIE is based on a BCC one can construct a
structure that contains both ABIEs and BBIEs at the same hierarchical
level. Is this correct? This does appear however to be illogical from a
UML point of view since an attribute can't be a class.</Michel>

Message Assembly is out of the scope of CCTS. Disregard the Message
Assembly box and the Assembly Component in Figure 4-2. A change request
has been forwarded to remove these artefacts in the next version. But
even the figure doesn't tell you how BBIE's and ABIE's are mingled. In
an Assembly spec it could (and should) be prescribed that BBIE's can
only be used as part of their ABIE.

<Michel>I would appreciate a definition and example of datatype as it
appears to be used all over the place with different meanings. My
understanding is that it is a set of values and thus only has
significance if these values are restricted. Is that correct?</Michel>

Not quite. If a Data Type without value restriction is needed it should
be defined. Then it probably will be unqualified and have the same name
as the Core Component Type it is based on. Whenever value restrictions
apply the name should be qualified.

<Michel>My understanding of the CCT is that it corresponds to a UML
attribute type and that a BCC corresponds to an UML attribute and
finally that an ACC corresponds to a UML class? Is this
correct?</Michel>

Yes.

<Michel>If this is so, then I assume that the ABIE simply inherits the
ACC through a generalisation and that the generalisation itself provides
the business context. Is this exact?</Michel>

Rather through Specialisation (but you may mean that). The context in
which the Specialisation applies should be defined as meta-attribute to
the ABIE.

<Michel>However, if this is so I therefore have a problem with the BBIE,
where does this fit into the UML picture?</Michel>

BBIE's are specialised from the BCC's. This is not a native UML
mechanism, but it is allowed and can be defined in UML.

<michel>I assume that the datatype notion can be derived from a
constraint on the type using a UML constraint built in OCL. Is this
exact?</Michel>

Yes, I think so. OCL is not mentioned or mandated in CCTS, but I am
advocating its use for this purpose.

<Michel>I know I'm asking an awful lot of questions, but since you
people have been working so long with the CCTs and UML you have
certainly worked all this out already and thus will save me the time in
trying to figure out the relationships.</Michel>

Sure, I am glad all work that went into CCTS has not been in vain and
that the CCTS is used in practise!


Regards

Fred


<<attachment: winmail.dat>>



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index]