[FOM] Object-Oriented Formal Mathematical Languages

charles silver silver_1 at mindspring.com
Wed May 5 13:11:49 EDT 2004


Steven wrote:
> In support of Tony Hoare - I worked with his group at Oxford while
defining one of the languages based on his formalism CSP - that
language was the Occam programming language that deals with
communication and concurrency - and we were very focused on the formal
verification of computer programs.

> It is only a matter of economics that lets software companies get
away with the liberal informality of modern programming languages.  If
the penalty for shipping faultly code was equal to the penalty in the
semiconductor business (which can amount to million of dollars) then
you would see software houses demanding formal practice, verification,
demonstration of certain properties (such as deadlock feeedom) and
proof that a program met its specification.
>
> As it is th cost of poor practice is simply passed to the consumer.
>
> That computer programmers are not mathematicians is matter only of
>economic tolerance and engineering pragmatics.

    I once worked for a company that produced
dBASE clones. To develop our products, it
was useful to look at dBASE bugs.  The company
owning dBASE (Ashton-Tate) published 78
pages of *known* dBASE bugs for the then
current version--and of course, we found
many that weren't published. The CEO of
the company I was working for used the
78-page list of dBASE bugs as a kind of
standard for determining when it would
be acceptable for market release of our
competing products.   (I remember being
shocked at hearing this.  Some of the
dBASE bugs were funny: a certain
type of line of dBASE code would
not run properly when the number of
characters in the line was evenly divisible
by four.)

Charlie Silver





More information about the FOM mailing list