However,
pro?¬?les offer a limited extension mechanism since totally new types can??™t be added to
the language. Also, pro?¬?les can??™t allow taking anything away from the UML since
pro?¬?les are based on existing UML concepts. Therefore the use of models for code
generation, analysis, checking, or documentation needs to access the extended
language concepts via mandatory and possibly unnecessary UML concepts.
Currently, the OMG standards give an indication that such removals could be
done, but no reference implementation or tool support yet exists. An obvious test for
such a standard would be to remove the whole UML with pro?¬?les and build a totally
different language.
Pro?¬?les can still be used in cases where the difference from basic UML concepts is
small. This leads again to applying the abstractions of UML and making any larger
deviation??”to map modeling concepts more closely to the problem domain??”would
lead to unnecessarely large and complex language de?¬?nitions. Usually, it is far better to
just say ???dog??? than to say ??????cat??™??™ then explain how???dogs??? differ fromcats.You may test
this yourself by de?¬?ning pro?¬?les to implement the sample languages from Part III.
Because of these limitations, the OMG has proposed another form of customizing
modeling support, the Meta-Object Facility, MOF.
Pages:
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131