Brief report
ABSTRACT
The
report gives account of the IFIP Working
Conference in program specification and
transformation
(Bad Tolz, FRG, April 15-17, 1986)
attended by 65 specialists from
different
countries including the report’s
authors, as well as information on the
Munich CIP project for
program transformation obtained during
the authors’ visit to the Informatics
Institute of Technical
University of Munich on the 14th
of April of 1986.
SUBJECT
MATTER
The
Conference
The
conference was being held in April
15-17, 1986, in the town of Bad Tolz,
FRG, 85 km from
Munich, and was devoted to an urgent and
actively developing direction of
present-day
programming – program specification
and transformation. The main goal of
works in this direction
is to increase the automation level when
constructing reliable and efficient
programs, to learn to
construct programs and proofs of their
correctness with substantial aid of a
computer. Here an
important role belongs to formalization
of the notions of “program” and
“problem formulation”
leading to algorithmic languages and
specification languages. Programs and
specifications
transformations are the tools that, once
having been materialized in software,
are being used for
automation
of the process of transition from
pre-algorithmic problems formulation to
reliable and
effective
programs for their solution.
The
conference was attended by 65 well-known
scientists from 16 countries – FRG,
USA, France,
England, Italy, Norway, Holland, Sweden,
Finland, Spain, Austria, Belgium, Canada,
Japan, and
USSR. Among them were developers of
experimental program transformation
systems, leaders of
programming
projects, specialists in programming
methodology.
A
series of talks by members of the
Informatics Institute of Technical
University of Munich were
devoted to CIP project carried out in
this institute. This project is aimed at
the development of
transformational programming methodology
where program development is viewed as a
process of
sequential transformation of
specifications and programs. A prototype
system CIP-S was created
allowing carrying out such
transformation process staying within
the limits of a “wide-spectrum”
CIP-L language and in many respects
relying on the aid of computer. At that,
transformations
application is done automatically
whereas selection of necessary
transformations from the vast
system catalogue as well as proving
their applicability conditions remains
mainly the human
programmer’s part of the job. The
paper by Bernhard Muller and Helmut
Partsch presented the
objectives of CIP-S system, its main
design decisions taken at the formal
system specification stage,
as well as the experience obtained by
application of algebraic approach to the
system’s specification
and transformational approach to the
system’s software development. The
paper by Rudolph
Berghammer formulated the main theorems
and transformation rules constituting
mathematical
basis for the notions related to the
notion of functions’ composition in
applicative language. The
paper by Bernhard Muller and Otto
Paukner formulated algebraic laws for
operations over sets. The
use of these laws as transformation
rules allows to get rid of some
specificational constructs and
convert them into algorithmic ones. The
paper by Bernhard Muller described a
language for writing
algebraic specifications using
higher-order operations and formulated a
formal theory of algebraic
types.
Some
papers presented at the conference were
concerned with considering the
experience in
implementation of concrete experimental
systems for program transformation,
generalization and
classification of the methods used.
First of all it is necessary to mention
the paper by Martin Fizzer
(USA, Institute of Information Sciences)
presenting a survey and classification
of up-to-date
approaches and techniques of program
transformation. The paper by James Boil
(USA, Department
of Mathematics and Informatics of
Argonne National Laboratory) described
an experience in
program development with the help of
TAMPR transformation system. Input
specifications for this
system are written in LISP and efficient
resulting programs are generated in
FORTRAN for
sequential and parallel computers. The
paper by Hiroshi Nakagawa (Japan,
Yokogama National
University) described a program system
for Prolog programs transformations
directed at
improvement of their efficiency and
based on generalized fold and unfold
operations.
A
great interest of participants was
evoked by the talk by Professor C. Hoare
(England, Oxford
University) formulating the means for
algebraic specification of communicating
sequential
processes and analyzing algebraic
properties of such specifications. Some
more presentations were
also devoted to specification means, in
particular the papers by Eugidio
Astesiano (Italy, University
of Genoa), David Vail (USA, Institute of
Information Sciences), P. Pepper (West
Berlin). It was
noted that algebraic specifications
allow writing problem formulations in
compact form, and their
algebraic properties make it easier to
prove correctness of generated programs
and correctness of
transformations.
The
authors of the report also delivered
talks presenting their results on the
subjects within the
conference scope. A.P. Ershov presented
the paper “Controlled Mixed
Computation and its
Application to Systematic Construction
of Language-Oriented Analyzers”, and
V.K. Sabelfeld
presented the paper “How to Consider
Properties of Operations in Global
Program Analysis”. The
presentations raised questions and
interest of the conference participants.
In
addition to participation in the
conference the authors seized the
opportunity to visit the
Informatics Institute of Technical
University of Munich. During the short
visit discussions with the
following members of the Institute took
place: Walter Dosch, Rudolph Berghammer,
Bernhart
Moeller, and Mrs. K. Samelson. In these
conversations the authors were
introduced to the current
state of CIP project and perspectives of
its future development and also watched
the demonstrations
of CIP-LS compiler operation and a
dialog library card-processing machine
running on Siemens
computer.
Discussion
of scientific and technical cooperation
issues
Besides
scientific discussions on the conference
topics the work aimed at establishing
scientific
contacts and identifying possible forms
of cooperation in the fields interesting
to Soviet specialists
was carried out. Positive results of
these talks may be summarized as follows.
-
Prof.
F.L. Bauer (Technical University of
Munich). A basic agreement was
reached on
handing over the CIP project
software for carrying on coordinated
experiments.
-
Prof.
D. Bjorner (Danish Informatics
Center, Lyngby, Denmark). A basic
agreement was
reached on holding an international
seminar in Denmark and inviting five
Soviet specialists
at the expense of the receiving
party.
-
Prof.
K. Koster (University if Neimen,
Netherlands). Proposals were put
forward on
delegating a Soviet specialist for
scientific work at the University of
Neimen on contract
basis.
-
Prof.
A. Haraldson (Linkoping University,
Sweden). Confirmed his interest in
inviting a
Soviet specialist to the university
on contract basis.
-
Prof
R.Dewar (New York University, USA).
A prominent scientist in the field
of
supercomputer software. Showed
interest and desire to make a
scientific trip to the USSR.
Propaganda
and counter-propaganda
The basis of propagandistic work
consisted in scientific advances of
Soviet scientists in topical
fields of informatics. In particular,
the talks by Soviet participants brought
appreciation and
interest that showed in proposals for
scientific and technical cooperation.
Foreign scientist
showed a great deal of interest in the
USSR state program of school
computerization and
scientific foundations of the basic
course in informatics worked drawn up by
Soviet scientists.
Informal discussion of
Soviet-American relations and Strategic
Defense Initiative allowed to
concentrate attention of foreign
scientists at negative aspects of the
USA administration’s policy
consisting in infringement of free
creative work, politization of science,
restricting scientific and
technical cooperation along with
strengthening American hegemony in
informatics and
computer engineering.
CONCLUSIONS AND RECOMMENDATIONS
-
Transformational
methods in programming remain the
central and rapidly developing issue
of theoretical informatics.
-
Although
in some directions Soviet works are
up to the world level or even set
the latter
(program schemata theory, logical
programming and program synthesis,
mixed
computation), an overall scale of
these works in the USSR and
especially their rate of grows
is still inadequate.
-
There
comes a time for creation of
transformational programming and
program synthesis
demonstration systems allowing to
get down to the matters of practical
efficiency and
technological implementation.
-
It
is necessary to undertake efforts
for more profound analysis of CIP
project and familiarize
oneself with its software as far as
possible in the scope of the
outlined scientific and
technical cooperation.
-
It
is very important to use the given
opportunities for long-term
scientific visits of Soviet
specialists to the “small”
European countries’ universities
with a high scientific image.
Active work in these scientific
groups will be a positive
contribution to scientific and
technical cooperation and will allow
to make up for subsiding of
Soviet-American scientific
exchange.
The
head of the delegation,
Academician
A.P. Ershov
Íàçàä
|
