resurrecting the Intel iAPX-432 32-bit microprocessor

[Eric Smith]

On Wed, Jul 22, 2015 at 8:02 AM, dwight <dkelvey at hotmail.com> wrote:
> I recall trading a board with one of the 432 family parts with you
> years ago for a disk drive unit.

Thanks again!  I've lost track of which 432-related hardware I got
from whom, but I remember that some of it came from you.

> I recall that everyone was thinking the military would switch to
> ADA

The part of the DoD that caused Ada to be brought into existence and
mandated its use was focused on long-term cost and reliability
benefits.

Unfortunately for any given DoD contract, the contractor could say "we
can get this done sooner and for less up-front cost if we do it in
<non-Ada-language>", and get a waiver. It wasn't long before all DoD
contracts were getting such waivers, so eventually they dropped the
requirement entirely.

> and the 432 was groomed for that.

Intel marketing claimed that the 432 was designed for Ada. That's not
actually correct. Intel had their own language in development, which
in early form was called Prototype System Implementation Language
(PSIL).  When Ada came along, everyone believed it was the future of
software, and that it was better to adapt to Ada than to try to push a
proprietary programming language. They found that Ada would map
tolerably well to the 432 architecture, though Ada didn't provide some
capabilities that were necessary to the operating system, so Intel
actually had an extended subset of Ada. Some features of the early 432
architecture were actually dropped, including coroutines, because they
couldn't be supported in Ada. The DoD Ada validation requirements
disallowed language extensions, so Intel sometimes referred to the
extended Ada as SIL (System Implementation Language), though it used
the same compiler and I don't recall there being any way to disable
the extensions.

> It was believed that more correct software could be created that
> way. At least the generals bought it.
> What wasn't known or thought about at the time was that the
> more serious errors in released code was a misunderstanding
> of what the code was suppose to do and not so much how it was
> doing it( at least, once you remove memory leaks  from the list ).

Actually it's pretty well documented that programming in type-safe
languages does result in more reliable code (as compared to e.g. C).
It doesn't prevent all bugs, obviously, but it catches at compile time
several classes of bugs that can only be caught at run time in
non-type-safe language.

> I'm curious, does anyone program in Ada?

Even though the DoD doesn't require it any more, it actually does see
a fair bit of use in the aerospace and telecommunication industries.
When companies are building software for themselves, rather than for
the DoD, sometimes they are willing to factor the long-term benefits
into their decision-making.

Boeing uses Ada for the flight control software of the 777 and 787.