Rational R1000-400 and RTEMS?

[Joel Sherrill]

On Wed, Nov 6, 2019 at 3:23 PM Chris Johns <chrisj at rtems.org> wrote:

> On 7/11/19 1:23 am, Per Dalgas Jakobsen wrote:
> > Last week we succeeded starting up an R1000-400 and have a working
> environment
> > on a FACIT A-4600 monitor.
>
> Oh my that is amazing. I have not seen one of those since the early '90s. I
> worked on the hardware side of a project written in Ada developed on one of
> those boxes. The FACIT terminals are nice, I ended up with one for many
> years
> after the Ada box was switched off.
>

Impressive to get that running. I did some Internet sleuthing and that's
not easy. :)


>
> > It just occurred to me that RTEMS might have been a target for
> cross-compilation
> > from the Rational R1000-400 back in the '80-90s.
>
> I remember Rational provided their own run-time. The target hardware we
> had was
> a DY4 68020 processor with hardware floating point.
>

The DY4 DMV152 (I think) had an RTEMS BSP eons ago. It was a MC68020 board.


>
> > The R1000-400 is a machine intended for team development and maintenance
> of
> > large Ada systems:
> > https://en.wikipedia.org/wiki/Rational_R1000
> > https://datamuseum.dk/wiki/Rational/R1000s400
> >
> > A log of our efforts to get it running, with some picture can be found
> here:
> > https://datamuseum.dk/wiki/Rational/R1000s400/Logbook
> >
> > If there is anyone here that may have some history, stories or knowledge
> related
> > to the R1000-400, we would be very interested to hear about it.
> Especially if it
> > involves RTEMS of course :)
>
> I did not use the Rational box but I remember somethings. The run-time
> parts I
> saw had a tick and basic tasking and I think an interrupt pragma but I
> seem to
> remember the software had a lot of sleeps and polls.
>

That's about all that's required for an Ada run-time. One of the original
goals of
the Army sponsorship of the RTEMS Project was to be a cross-compiler Ada
run-time.
We did implement it for Tartan and Telesoft but not Rational (or any other
vendor).
It pointed out that the Ada products suffered from some of the same things
RTOS
products do that RTEMS was supposed to address. The run-time interfaces were
highly proprietary, subject to change, required an expensive (USD100K in
1991)
source access license fee which needed to be paid for each version, and
were
not really designed to be delivered as source code.

We had suffered from Ada vendors and hosting on a VAX 11/782 (dual CPU)
running
VMS when we heard of GNAT being developed. At that point, we made contact
and
began to develop POSIX thread support for the C RTEMS to be a run-time for
GNU
Ada. That is still working today.  We actually passed Ada validation for
GNAT/RTEMS on
a SPARC erc32 back in ~1995 I think.

There actually was an implementation of Ada in parallel with the C version
but the
C version was more more popular. We could never proof anyone actually used
the
Ada version and effort moved to where there were users.

The 80s and early 90s were an interesting time for workstations before
things became more
homogeneous. I recall going to a premiere event for an SGI  workstation
(can't remember
which one), but that was a nice looking machine. Style and uniqueness seem
to have
disappeared.

--joel


>
> All best with project.
>
> Chris
> _______________________________________________
> users mailing list
> users at rtems.org
> http://lists.rtems.org/mailman/listinfo/users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20191106/dac41642/attachment.html>