Meeting on Saturday

[Andrzej Tarlecki <tarlecki@mimuw.edu.pl>]

Dear Colleagues,

Just a word to remind you once more about the meeting of our
methodology subgroup of CFI in Oxford, on Saturday 16th March, 14.00,
at Lady Margaret Hall. For more details on the location and all local
arrangements see earlier messages from Grant Malcolm; if you intend to
come but have not registered for the meeting yet, please contact him
immediately to check whether it is still possible to arrange some
accommodation etc.

As for the plan for that meeting: so far there is nothing very
specific. As you know, this is our first meeting, so we will have to
discuss some very general issues:

1. What is exactly the scope of the subgroup; is it sufficiently
reflected by the scope and aims as formulated earlier; what exactly
should be the outcome of our work, etc.

In my view, the aims and scope formulate the right direction (after
all I have proposed them to much extent :-) although perhaps are not
quite specific enough to be directly used as a work plan.

2. how should we proceed with the work, how to split it up, etc.

I presume, at least in the beginning this will proceed much as is does
in the language design subgroup now, by technical meeting and
production of preliminary technical notes on various topics, with the
hope that a unniform, consistent picture will emerge.

In fact, many of the notes written so far for the language design
subgroup are closely related to the methodology issues, so maybe it is
worth having a careful look at them as contributions to our work
before the meeting.

Most roughly --- and without having thought about this myself too much
--- I envisage that the methodology shoul deal with two major areas:

A. Methodology "in the small": how to develop "small" specifications,
mainly of specific data types (stacks, queues etc). Here would come
the discussion of constraints (initiality, reachability) and their
lack (looseness, nondeterminacy); constructors and observers of a data
type; (relative) completeness of specifications; etc.

B. Methodology "in the large": how to specify and develop "large"
software systems, with their (structured) specifications. Here we
would deal with the general view of software development process
(refinements etc); role of behavioural interpretations of
specifications (stability?); designing the structure of the system to
be developed (modules, parameterisation); backtracking in developments
(reuse of components, flexibility of specifications, etc); one more
etc to indicate that the list is far from being exhaustive :-)

I believe that above topics can be discussed and presented in a rather
formal, mathematically well-founded way. Thera are also a number of
other issues, like software life-cycle, semi-formal methods in our
methodology, etc, which perhaps should be in the scope of our
interests as well.

One major topic to attack soon, as it may well influence the work on
the language design, is the general picture of a relationship between
specifications and programs, where the latter may be understood on one
hand as "real programs" in a programming language, and on the other
hand as "algebra definitions". The CFI formalism will cover the world
of specifications, of course, but to what extent it should also
include the world of "programs"? How much of refinements and
development should be expressible in the language, and how much should
be left to some extra "environment" (whether formalised and
computer-supported or not)?

All for now, more in Oxford.

Looking forward to lively and effective discussion in Oxford,

With best regards,

Andrzej