The current (Oct 99) issue of Communications of the ACM has an
article "UMLoquent Expression of AWACS Software Design" in which
the authors talk about UML <-> Ada and say things like "There is
currently no known means ..."  and "We know of no means..."
about generating Ada from UML or UML from Ada.  The fact the
authors know "no means" may reflect their own limitations, but
will probably be taken by readers to mean "Ada and UML don't work
together".  Could those who know this subject please comment.