brainerd@unmvax.unm.edu (Walt Brainerd) (10/05/89)
To x3j3 and comp.lang.fortran readers (sorry if you are both!):
Something like the following will appear in the next issue of the
Fortran Journal. I thought perhaps you folks might like to see it first.
Between official sessions of the Lahey Fortran 8x Implementation Symposium,
several people were discussing the issue of creating a validation suite
for Fortran 8x. This is a project that probably will require several
person-years of effort and, for obvious reasons, it is not likely that
a validation suite created by one Fortran implementor will be widely accepted.
More than one person has proposed some sort of "consortium" be created to
approach this problem and we at the Fortran Users Group and Fortran Journal
thank this might be an ideal project with which to become involved.
The FUG has access to people with extensive knowledge of the standard
and a large pool of people that can implement a validation suite.
What the FUG is missing (naturally) is financial backing.
We haven't thought about this long enough to come with a whole lot of good
ideas, but one possibility is to have the FUG be the organizers of this
"consortium". The members of the consortium would be those willing to
put up enough money to finance a significant part of the suite development.
The members of the consortium would participate in the design of the
suite, would have access to it at all stages of its development, and would
"own" a share of the revenues produced when the validation suite is used
by others. However, we would envision that the whole process will be open
in that others may see how the design is going and offer suggestions for
its design. The members of the consortium could be implementer or user
organizations.
A important component of this plan is to gain acceptance by NIST, BSI,
and other organizations with an official interest in the validation.
If other organizations are planning to do validation suites, we need to
get together, perhaps under the auspices of NIST et. al. to plan an approach
to the whole thing. We have talked with a person who appears to be the
right contact at NIST and he has expressed an interest in holding some
kind of workshop on this topic, probably early in 1990. This person is:
L. Arnold Johnson
N I S T
National Computer Systems Laboratory
Building 225 Room A-266
Gaithersburg, MD 20899
If you have any comments, ideas, or just want to be informed of developments,
please contact Mr. Johnson or the FUG (by e-mail to the sender of this msg).
--
Walt Brainerd Unicomp, Inc. brainerd@unmvax.cs.unm.edu
2002 Quail Run Dr. NE
Albuquerque, NM 87122
505/275-0800