Re: Why not MaRTE was Re: An OS in Ada, why not RTEMS ?

[Nick Roberts <nickroberts@blueyonder.co.uk>]

On Tue, 30 Apr 2002 10:49:52 +0200, Alfred Hilscher 
<Alfred.Hilscher@icn.siemens.de> wrote:

> > I have had time to "revisit" the EduOs-Ada
> > project and during much needed research run 
> > accross RTEMS.
> > 
> > A brief look at the RTEMS specs would suggest 
> > that RTEMS either meets most of the needs 
> > discussed in 2 or 3 years (at least) of threads 
> > in cla or is capable of supporting those needs.
> > 
> > I am wondering why the contributors to the said
> > threads do not regard RTEMS as a suitable OS to
> > adopt as a platform for playing with ?
> > 
> > Any insights appreciated, thanks.

Alfred,

If you are talking about the requirements for AdaOS 
(and I'm not at all sure you are), then (to my mind) 
RTEMS falls short of far too many of our 
requirements.

Chiefly, RTEMS has absoloutely no security 
provision. Following from that, it has no provision 
for (enforced) resource management, and all the many 
concomitant facilities.

In my detailed evaluation of RTEMS, I didn't get 
much further than that. There did appear to be a 
number of other flaws and deficiencies. RTEMS may be 
excellent as a real-time OS/RTS, but I think for 
AdaOS purposes, it's a dead duck.

There are some requirements of the AdaOS kernel 
which, it needs to be appreciated, all or most 
existing microkernels/RTOSes will not vaguely meet. 
For example, AdaOS will be distributed, supporting 
the arbitrary migration of most processes between 
workstations; that requires a very special set of 
kernel characteristics for a start.

I repeat, all the foregoing applies ONLY IF Alfred 
was referring to AdaOS at all.

-- 
Nick Roberts