[net.lang.ada] "Question about CAIS recommendation" from larry@Jpl-VLSI.ARPA

munck@MITRE-BEDFORD.ARPA (Bob Munck) (06/18/86)

Larry,

> my suggestion is to include in the SDE Request for Purchase ... a
> requirement for the ability to easily upgrade to the future CAIS.

The problem is that such a statement in an RFP is pure trouble.  You're
requiring the bidders to state how and at what cost and schedule they will
do an unknown thing.  By "unknown thing" I DON'T mean the upgrade to the
future CAIS, but the creating of the "ability" to do so.  Depending on
the wording, this will inspire highly creative and completely
content-free verbiage in the proposals; depending on the contract terms,
it may give the chosen contractor a license similar to that which James
Bond has.  On the other hand, we do this kind of thing ALL THE TIME!

I attended the same session in Houston (with a pocket tv for the
Celtics) and felt that there was one fact that was so obvious that it
didn't need to be stated.  That is, the KIT/KITIA has been working for
about half a decade toward meeting the same general requirements that
the NASA SEE (SDE?) must fulfill.  The CAIS 2 definition might not quite
meet the schedule, but it's closer than anything else.  Certainly the
best bet for SEE is to go with the CAIS rather than start over.  The
definition will be in substantially final form when NASA needs it, just
not completely through standardization.  Heck, I'm working NOW on a
CAIS 2 implementation for the 80386.

BTW, I enjoyed the vehemence with which the NASA folk rejected any use
of UNIX, both (I assume) the NJ original and mutations from east and
west.  They do understand the difference between a system that lets
individual hackers and gurus churn out code and one that supports
development of good software by teams of programmers.
                     -- Bob Munck, MITRE