In tools of this type the graphical metamodel was
made with the same modeling tool functionality as the resulting tool, but using a
special graphical metamodeling language (whose metamodel had been bootstrapped
from a textual description, and often had special links to tool functionality reserved
only for that language). The resulting metamodel was ?¬?rst transformed into a textual
language, which was then compiled as above.
This separation of metamodeling and modeling had an important drawback: it was
not possible to interactively test the results of metamodeling immediately, because of
the long transform??“compile??“link??“run cycle tomove from metamodeling to modeling.
This was a signi?¬?cant hindrance to the metamodeling process, as if a problem was
spotted while testing a modeling tool, the user had to exit the tool, restart the
metaCASE tool, read in the metamodel, edit it, transform, compile, and link it to form
the resulting modeling tool. These steps often required separate commands, manual
text editing and external tools. The resulting modeling tool could then be restarted,
and the existing models generally needed to be explicitly updated to use the new
metamodel before the change could be tested.
Belowwe shall take a look at three tools of the 1990s. Others of course existed, and
some tools born in the 1990s are still going strong (we will see current tools later in the
chapter).
Pages:
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688