Each model may contain further links to other models. If these submodels are in the
same modeling language, the generation for them will most likely be a recursive call
to the top-level generator, but nowon that submodel. If the submodels are in a different
modeling language, they are somewhat more likely to be including information by
reference rather than containment. The generator may then simply pick up the
information it needs, or generate a reference, without needing to process the full
content of those submodels. The full content will be generated via some other path
from the top-level models.
In some cases, the information partition that is best from the point of view of the
problem domain may be at odds with that required by the implementation domain.
The implementation language or frameworks may require a certain set of information
in a ?¬?le, but that information might be spread over several models in various modeling
languages. In that case, it might make more sense to have autobuild call a generator
per ?¬?le or per ?¬?le type, and have that generator call partial generators for each
modeling language it needs to visit for information. Generally, though, the pattern is
that a generator per modeling language will contain a number of subgenerators, one
for each ?¬?le type to be generated.
Pages:
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560