Re: Is Linux right for Embedded?

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

If anecdotes count for anything, I could enumerate a long list of hours
spent in the last few weeks (We're busy trying to test quality into the
system! :-) where I've been tracking down memory leaks, dangling pointers,
illegal memory references, and bad array indexes all because of pointer
usage and/or lack of checks in C which - FWIW - wouldn't happen in Ada
because of the a) limited use of pointers, b) scope rules, c) compile time
checks or d) run time checks.

I recall a Bell Labs study of bugs in one of their big switching
applications in which the report found a whole slew of common bugs which
were classified into types & examples given. Their recommendation was to
institute coding standards and code reviews to check for those specific
problems. The overwhelming bulk of the bug types they identified were
impossible to commit in Ada because of compile or runtime checks and the
rest were highly unlikely because Ada doesn't depend on addresses/pointers
for everything in sight.

Its anectdotal, but it adds a level of experience and reasoning that
explains *why* the studies indicate superior productivity/error rates with
Ada vs C.

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/


"Ted Dennison" <dennison@telepath.com> wrote in message
news:gGHu7.14498$ev2.23688@www.newsranger.com...
>
> However, if there *were* numbers on C vs. Fortran, I wouldn't be shocked
to see
> Fortran (even F77) come out on top. C's overeliance on pointers cause a
great
> deal of its problems. But this is all theoretical. The numbers on Ada and
C
> actually exist.
>