Having now read the CACM article, "UMLoquent Expression of AWACS Software
Design", I must add my apologies to the authors for my earlier ill-informed

The statement in the article, "Attempts to forward engineer classes
in NMT's logical view into Ada specifications have been of little value
of the fuzzy semantics associated with mapping classes represented with the
into Ada", is not a denigration of Ada.  _Out_of_context_ one might get the
that the Ada language itself is in some way inadequate, or so "off the
track" that UML doesn't map well to it.

That is clearly _not_ what the article is saying, the criticism is in fact,
believe, that the UML mapping and available toolsets inadequately exploit
Ada's representational capabilities.  As an example the article cites
difficulties in getting the toolset to map UML classes as child packages,
in forward and reverse engineering.  And in my opinion, child packages are
one of the very best features incorporated into Ada 95, so the fault here
within the UML sphere.

Marc A. Criley
Software Architect
Lockheed Martin M&DS
(610) 531-7850
[log in to unmask]