Re: CFI - Meta-Requirements and Meta-Design

[Joseph.Goguen@comlab.ox.ac.uk]

I have some brief comments on the meta-requirements for CF:

1. To the "Overall" list, I would add: the CF should be developed as modularly
as possible, to allow further incremental development.

2. Under "Functionality": I believe that the bullets relating to requirements
should not be directly supported by a CF language, but rather should be part
of a functionally separate *environment* supporting use of the language.  

3. In particular, development specs from reqs, relations between reqs & specs,
between specs & imps, and adjustments of reqs, specs & imps all have to do
with support for software evolution, which should involve dynamic links among
all the various kinds of software objects, with appropriate configuration and
version management.  F. Pinheiro and I recently finished a paper on how to
provide such support in a hyper-media (HTML) environment implemented in FOOPS,
an algebraic declarative object oriented spec & programming language.  We can
provide copies to those who may be interested.

4. Support for evolution will be crucial to the large scale practical use of
any kind of formal method.