The symbols for the new modeling
language were de?¬?ned in a separate graphical symbol editor. The editors could
generate these two parts of the metamodel in the textual metamodeling language, and
the user would join these parts by hand to form one ?¬?le. This ?¬?le could then be run
through the Moffer metamodel compiler to generate the binary metamodel ?¬?le that
drove the generic modeling tool part. If a metamodel was changed, existing models
based on it could be explicitly updated to use the new metamodel: the actual update
was performed automatically. The ease and incremental nature of metamodeling were
recognized in the tool comparisons as the main advantages of MetaEdit.
MetaEdit had several limitations, mostly in its modeling tool functionality. Each
model ?¬?le was limited to a fewtens of diagrams, all of the same type, and there was no
362 TOOLS FOR DSM
linking between model ?¬?les. This prevented proper implementation of most modeling
languages containing several integrated diagram types. An odd implementation of
relationship metamodeling required creating several near-duplicates of relationship
types in some not infrequent cases (Kelly, 1995), also complicating modeling use with
extra types. The conceptual??“representational distinction was rather weak: each model
?¬?le was conceptually one large graph, of which subsets were visible in different
diagrams.
Pages:
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691