Re: Increased Interest In Ada?

["Marin David Condic" <marin.condic.auntie.spam@pacemicro.com>]

I have some small familiarity with GNAT/RTEMS - which is to say I basically
know what it does, but with no knowledge of how it actually does it.
Basically, this wouldn't be a bad way to go.; Let's say, for the sake of
argument, that one had a version of GNAT that ran on a PC and was targeted
to a bare board MC68040 processor. One would need to compile RTEMS to this
processor to have the runtime services needed by Ada for tasking, etc. From
there, you would compile your application using GNAT and link it with RTEMS
& then have an image that would be executable on this bare board MC68040.
(Someone correct me if I'm fundamentally wrong - details at this point don't
concern me...)

Key to this would be that you'd need a fairly sophisticated linkage editor
that ran on the PC and targeted the 68040. You'd need to be able to locate
chunks of code/data in specific memory locations because presumably you'd
have things like EEPROM and DMA devices, etc.You'd have to get out of it
some kind of load image in a usable format (ELF or S-Records or whatever).

From there you need some means of getting the load image into the bare board
machine. That means having a PROM burner or some kind of bootstrap code
loaded on the machine to pull data off of some port and start loading it at
addresses required. The bootstrap needs some kind of software at the other
end of the port to read the loadable image and feed it the bits & bytes in
some protocol it understands.

Now the important thing to note here is that this is the *BARE MINIMUM* one
needs *JUST TO GET STARTED PROGRAMMING*. (sorry for shouting!) You'd still
want things like a source level debugger, some kind of on-board monitor to
wrap with your application, probably some software packages that could be
used to shield you from device specifics (although that would obviously be
one of the student exercises, you'd want them to be able to work with them
only as required by the course & just have some stuff available for the
other devices.) We could obviously go on with wanting things - like a
JTAG/EJTAG interface with software at the PC end, etc. I can easily imagine
*wanting* a lot! :-)

While we're at it, we'd need real good documentation for the SBC and its
hardware devices because you couldn't expect students to just simply *know*
how to write code for Brand X A/D Converter and it would be a lot to ask
them to go start pestering vendors for data books. (Hell! It's a whole lot
to ask of us professionals - but at least we can justify it along the lines
of "If the company is too *stupid* to go get me the manuals I need, I'll
burn up their money surfing the net or calling the companies until I get
it.) Documentation would have to extend to the overall system as well.
Someone has to answer the question "How do I go from some source file with
the embedded "Hello World" program in it to code actually cycling in the
box?"

From there, we'd need to either find or write some kind of college level
text that addressed embedded programming from the level of all the things
that will go on in our little SBC. I have only run into one text that
addresses this at all ("Programming Embedded Systems In C And C++" by
Michael Barr) and it only addresses some of the things you'd likely see in
an embedded computer - and as is obvious, not from the Ada perspective. (If
you know of others, I'd be glad to hear about them. I'd like to see
something that dealt with things like A/Ds and the devices that live at the
other end of them.)

Now the problem as I see it is this: Nobody has all these pieces pulled
together all in one place using Ada (at a low price, at least), but it
*DOES* exist (mostly) for C and maybe C++. You can go to any number of
vendors who will sell you an SBC development kit that will plug into your PC
with all the appropriate software at the PC end, etc. You can be up and
programming the card with C in short order and maybe the only thing you're
really missing is the college level text. Pulling together all this stuff in
Ada is certainly feasable, but it would be a non-trivial amount of work.

Since great minds think alike, I'll agree with you that the PR value for Ada
would be high because it would demonstrate how easy Ada is relative to C in
this arena. I'd go one step further in saying that *IF* the kit were to
exist, a *LOT* of EE profs would be tempted to structure a course around it
because it would eliminate a ton of work for them - hence even more PR
value. Throw on top of it that every EE student who's first embedded
experience is Ada would likely go on to industry with a favorable impression
of Ada and start pushing for its adoption. And of course, if the card itself
were fairly generally useful, you've got a commercial market for it as well.

My only problem with this idea is that my full-time occupation is not the
development of such kits and as a speculative, part-time venture I just
don't think I've got the time to do it. (Not in any reasonable timeframe!)
Maybe a vendor or professor or idle-rich-kid (or several of them) might get
interested and start pulling the pieces together.

MDC

--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com
Web:      http://www.mcondic.com/


"Tarjei T. Jensen" <tarjei.jensen@kvaerner.com> wrote in message
news:97668g$718@news.kvaerner.com...
>
> Marin David Condic wrote
> >I'm going to examine the PC/104 thing a bit more thoroughly. If you think
of
> >any other possibilities for an SBC to which GNAT might target with
minimal
> >fuss, let me know... Thanks.
>
> Check with oarcorp. They have something about a GNAT/RTEMS combo on their
front
> page. That might be worthwhile to examine.
>
> I suspect that anything quick and easy for a student/hobbyist to do is
hard to
> find information on. It seems to be the Ada way. I think the PR value of
such a
> "kit" would be incalcuable because everybody could see that Ada is easy.
It
> would be ridiculously easy to check it out for yourself.
>
> Greetings,
>
>
>
>