Look how
occurrences of the concepts can be linked to form a whole product or feature, and
classify the different kinds of linkages to create the relationships.
As the number of concepts and relationships increases, think a little about how
you may want to reuse elements of a model in other models, and which clusters of
concepts and relationships have high cohesion and low coupling with other clusters.
This will give you hints as to whether you need more than one kind of graph, each with
its own set of concepts and relationships (possibly with some overlap). Try also to
think whether a model for a largish product or feature would likely become too large
to be manageable. If so, consider adding a concept that simply points to a submodel,
to allow model layering or a decomposition of a large model into a hierarchy of
smaller models.
Finally??”for the moment??”consider the variability between products, features,
and instances of concepts. What is it that needs to be recorded about these differences?
What extra information is needed to generate code, or to help the generator decide
which of several similar variant ways of doing something it should follow? Add
support to the modeling language to capture this information, for example in the form
of extra properties on the concepts, relationships, and models.
Pages:
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643