Re: Progress on AdaOS (Was: Re: How Ada could have prevented the Red

["Marin David Condic" <dont.bother.mcondic.auntie.spam@[acm.org>]

"Didier Utheza" <ddutheza@bu.edu> wrote in message
news:Pine.A41.4.10.10108201848570.29818-100000@acs5.bu.edu...
>
> Now, in the case of AdaOS. The project has, as you said, two major
> difficulties. The first one is to develop a C-independent compiler. This
> implies to actually develop several kernels to address the various
> processor (intel, alpha,...). The development of an Ada compiler is not
> given to everybody. Maybe a cooperation between the AdaOS team and the
> GNAT group could ease things a little bit. Everyone will gain from it, and
> the time spend on it would be less.
>
Don't know what you mean by a "C-independent compiler". An Ada compiler not
written in C? I don't think it makes much difference.

If it were my project, the first thing to do would be to put the project on
some sort of basis by which the developers stand to make some money off of
it. That would create some incentive to actually get something done.

From there, you just pick a compiler as an interim development tool. (Gnat
would probably do just fine for a time.) I can see why it would be desirable
for the project to have its own compiler eventually, but there are lots of
good compilers that could get you started with what the *real* mission is:
developing an OS. Target the PC architecture initially, but leave the door
open for everything else and work the compiler issue as a sideline effort.

> The other difficulty is of course to start developping the OS itself. As
> it was said in the thread, if you have to wait for the compiler before
> developping the OS, it will become a fuzzy dream (it may become it
> already). Now since we have GNAT, it is still possible to start something
> and the members of the team that direct the project will have to do  at
> some point the transition. The GNAT group may be interested (a scary

That's the point: If you're going to build an OS then build an OS! Stay on
target and don't get distracted by the things that are sideline issues. Get
some minimal, working kernel going and people may get more interested in
contributing as they have something to play with and enhance.


> thing is that c(++) real time system are appearing now - they reinvent
> the wheel!), they may not have the manpower for such a project but have
> the experience. Also, many dedicated and embeded AdaOSes exist and the
> experience is there. A question: what is the compiler used to develop
> these OSes?

Lots of them. It depends on the machine architecture you are looking at. In
the embedded world, you look at your hardware, then start shopping around
for a compiler that targets the hardware & generates the best code for your
problem domain. I believe RTEMS is compilable by Gnat, but should be
compilable (probably mostly - some tweaking is usually necessary) by other
compilers if you have a target that Gnat doesn't support.


> The people that say that the project needs to show some code directed to
> the OS are rigth. The problem is who is going to do it. Who has the time
> and the knowledge to embark in this project?
> So, here comes the main question: what happened to Nick Robert. A
> leader or leaders (people to keep the project on the move) is needed. Nick
> is the one that created the project andnobody seems to know what happens
to him (the administrator of the AdaOS
> site: Marcelo de Freitas doen't know himself.
>
Any project needs some strong leadership to point the direction, inspire the
troops and keep things on-track. An OS is pretty ambitious and I doubt that
it can be done by committee. If a project like this is to succeed, it needs
one (or a very small handful of) dedicated guy(s) with a vision of what it
is supposed to be working on it long enough to get the core off the ground.
Once the direction is set and some initial progress is made, others will see
the direction and work with it.


> At last, there are plenty of people (like me) that play with Ada and do
> not have a clue in system/OS programming but would like to help. With a
> little bit of directive a lot can be done on the doc, code
> duplication,...side.

No doubt anyone could be useful to a project like this, but as you observe,
most people need a little help in being pointed in the right direction. I
have nothing to do with the AdaOS project, so I couldn't advise you about
the goings-on in that group. However, if they did start rolling along and
you thought you'd like to be involved, I'd suggest asking what are the
things that need to be done and find some piece of it that you have *some*
understanding of and work on that. It can be a valuable learning experience,
if nothing else.


> *BSDs and Linux were able to take off because a limited number of people
> with the deep knowledge of OS internals gave to the following layer
> something to build upon. It looks that any free open project to develop
> through the web would have to follow this path.
> There are very good Ada name in the AdaOS project. The knowledge is here,
> something just has to be done with the coordination.
> That is the most difficult step in my opinion.
>
I'd like to see them succeed as well. Certainly, there is tallent and
knowledge on the project, but I believe that unless someone goes off and
develops some small starting piece of it and says "Here it is - lets build
on this" nothing much is likely to come of it.

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/