ylkingsbury@watdragon.waterloo.edu (Yvonne Kingsbury) (05/25/89)
ICR presents a colloquium on
Remarks on Applied Denotational Semantics
by Dr. Andrzej Blikle
of Institute of Computer Science, Polish Academy of Sciences, Warsaw, Poland
ABSTRACT
During the last fifteen years we have been witnessing the development of
three approaches to the mathematical modeling of software: denotational
semantics, algebraic semantics (abstract data types), and operational
semantics. So far, the most successful in applications seems denotational
semantics. This is due to its technical version VDM (Vienna Development
Method), which has been applied in several industrial projects including
the first ADA compiler developed in Europe.
The industrial success of VDM has stimulated further research in applied
denotational semantics and has contributed to the creation of several
European projects - such as e.g. RAISE or MetaSoft - devoted to the
development of denotational-semantics-based tools for software engineering.
These projects put more emphasis on its algebraic aspects, while abandoning
some traditional technicalities of denotational semantics such as reflexive
domains and continuations. They also raise some general methodological
questions concerning the application. Our talk is devoted to some of these
questions.
In the first place it will be argued that the denotationality of a semantics
of a software system is, in a certain sense, a property of the system itself,
rather than just of its description. We show that a software system with
denotational semantics provides an adequate ground for structured programming
and for structured inductive proofs of program correctness and that the
resignation of denotationality may lead to the loss of these attributes.
We also discuss the trade-off between denotationality and abstractness by
showing that the nondenotationality of a semantics may be always ``repaired''
on the expense of loosing some of its abstractness. This leads to an important
pragmatic question, to what extent and in which situations can we sacrifice
denotationality and/or abstractness of a semantics.
We also argue that the attributes ``denotational'' and ``operational'' are
orthogonal rather than opposite. This is illustrated by a short discussion
of a so-called structured operational semantics and of its relationship to
denotational semantic.
The listeners are not expected to be experts in the subject of the talk.
Some familiarity with many-sorted algebras may help, but is not necessary.
Major ideas will be explained on examples.
Wednesday, May 31, 1989
3:30 p.m.
DC 1302
Everyone is welcome. Refreshments served.