Re: [UML Forum] the meaning of class/dataType

[H. S. Lahman]

Responding to Safa...

As i read through UML specification that:

�class has a set of attributes/operations and
data type has a set of attributes/operations

1)with regards to attributes/operations of data type
could you please clarify their semantics ?
because i don't know how such a data type has attributes and operations?

The proper terminology in OO circles is abstract data type, or ADT. An attribute is an abstraction of knowledge; it simply identifies something that all entities in a class should know. Note that such a definition is independent of the computing environment.

Data, OTOH, are values and they have a concrete representation in computer memory. Generally the computer has a very generic set of valid types for data -- integer, real, character, etc. Using hardware-based data types ties one to specific computing platforms. If one wants to move the program to a different computer, one may have to modify the data types for the new computing environment. To get around that problem one can define ADTs in UML that abstract the data in problem space terms so that the model is portable without change across platforms. ADTs also allow one to provide types for knowledge values that are more relevant to the problem space -- money, volume, length, count, etc. This allows much greater discrimination among types than computer hardware provides, which is useful for thing like automated error checking (e.g., detecting adding a money value to a length value). One can then map the ADTs to hardware storage types when actually implementing the model.

However, since classes only define what their member objects must know, there is a subtle difference between a class attribute and data -- the attribute knowledge does not need to be stored; the object could compute the value of the knowledge whenever it is asked for it. Thus an ADT can also have operations when it is actually implemented. (It can have other operations that are very convenient, such as synchronizing values of the same attribute in different subsystems when a value is modified.)

Thus the best way to think of an ADT is that it is a type mechanism to map knowledge attributes to specific implementation environments. ADTs are typically defined externally to the UML model (i.e., they are part of the MDA transformation rules used to transform the UML model to a different model, like a C program). Thus the transformation rule might say that a 'money' ADT in the model must be stored as a BCD value in memory. When one wants to port the model to a different environment, one just needs to change the ADT definitions (at most) and doesn't need to touch the model itself. Thus ADTs characterize class attributes, but they are quite different things.

2)What�s the main differences between class and datatype?

At the Assembly level, it is almost impossible to tell the difference. When UML models are transformed to non-OOPLs, like C, via a code generator there is very little difference.� When UML models are transformed to an OOPL, like C++, ADTs will be explicitly defined in header files for the mapping to storage types and explicit getters/setters will be provided operations.

Basically a class defines a set of objects in terms of the unique suite of problem space knowledge (attributes) and behavior (operations) every member object must have. ADTs qualify the knowledge with an abstract type that will be used to map between the problem space ADT in the model and the hardware.

-- 
Life is the only flaw in an otherwise perfect nonexistence
   -- Schopenhauer

Imagine how much more difficult physics would be if electrons had feelings
   -- Richard Feynman

Rene Descartes went into a bar. The bartender asked if he would like a drink. Descartes said, "I think not," and disappeared.

H. S. Lahman
H.la...@verizon.net
software blog: http://pathfinderpeople.blogs.com/hslahman/index.html
software book: Model Based Development, Addison-Wesley, 2011
geology book: The Evolution and Utilization of Marine Resources, MIT Press, 1972