Re: Ada Operating System

["John R. Strohm" <strohm@airmail.net>]

"David Starner" <dvdeug@x8b4e53cd.dhcp.okstate.edu> wrote in message
news:a76lbh$8i42@news.cis.okstate.edu...
> On Sun, 17 Mar 2002 19:25:09 -0600, James Ross <rem.jr@rem.webross.com>
wrote:
> > Sure, sounds like bizarre things to do -- but
> > the OS should not allow it regardless what reason I might have for
> > doing it.
>
> I _hate_ when gratitious restrictions stop me from doing what I need to
> do. When the user really, really wants to shoot themselves in the foot,
> the system's job is to provide the gun. (Note that Ada follows this rule
> - machine insertions, unchecked deallocation, unchecked conversions.)

The simplest answer, which is generally unpopular with the C crowd, is that,
to a first approximation, 99.997% of the time, the user does NOT want to
shoot himself in the foot, and reasonably expects the system to KEEP him
from shooting himself in the foot.  The appropriate answer is to provide
mechanisms that will catch him, 99,997 times out of 100,000.  At the same
time, you provide an override mechanism, preferably about three to five
levels deep in "Are you sure?"-equivalents, to WARN him that he is about to
lose toes and blood and ruin his shoes.

The C crowd seems to believe that, because you need this mechanism 3 times
out of 100,000 tries, it must be easily and immediately available 100% of
the time, and the user is required to look out for himself the other 99,997
times.  I have a problem with this approach.

> What if I'm about to rearrange some partitions, so I need to move some
> data onto the only stable partition, which happens to be /usr/lib?

I guess the obvious question has something to do with why your system has
partitions that require periodic manual rearrangement.

Remember, there is NO excuse for poor design.