Re: Linux Kernel Security - Can it ever be 100%
On Sat, Dec 06, 2003 at 02:39:38AM -0800, Tom wrote:
> On Fri, Dec 05, 2003 at 10:55:23PM -0700, Paul E Condon wrote:
> > > [This rant is probably full of shit] :-)
> > >
> > Yes. ;-)
>
> The reason it sounds so dumb is everybody would be out of a job.
> I've never met a single person yet who can truly look at statements like
> this and say "yes, let's do that."
>
> 20 years from now somebody will suggest the same thing and win a nobel
> prize.
>
The person who wins to Nobel is not always the person who did the
work. The work that is recognized by the Nobel award is not always
truly important.
The initial rant was about making computer software conform to a
specification. This is not a really new idea. It suffers from two
problems that remain unsolved: 1) all statements, including
specifications, that are written in a natural language are inherently
ambiguous. 2) Proving that an algorithm implements a particular
specification correct has been shown, by experience, to be very
difficult.
I think there is a possibility of new ideas on this topic. This rant
does not appear, to me, to contain anything new. It might have been
appropriate as part of a research proposal 20 yrs ago, but today it
doesn't cut it.
I am 71 yrs of age this month, and retired. I am not worried about
being put out of a job. Debian, GNU, and Linux are putting people
out of jobs, by being the low cost provider of quality software.
(Flame bait ;-)
In fairness to the original ranter, the ideas are not rediculous.
But they have been considered. If one hopes to make progress one
should be aware of prior work, and attempt to address the reasons for
its failure to solve the problem.
--
Paul E Condon
pecondon@peakpeak.com
Reply to: