Re: [avr-gcc-list] "Volatile"

[Colin Paul Gloster]

On Tue, Apr 19, 2005 at 08:53:52PM -0700, stevech wrote:
"My Two Cents:  Operating system functions like atomic access and mutual
exclusion and multitasking do *not* belong in a programming language
standard. The DoD tried it years ago with Ada and it was a huge failure."

Had multitasking in Ada been a failure? If so, had this been because of
inclusion of multitasking in a language as opposed to an operating system?
In HENRI E. BAL, RAOUL BHOEDJANG, RUTGER HOFMAN, CERIEL JACOBS, KOEN
LANGENDOEN, and TIM RšHL, "Performance Evaluation of the Orca
Shared-Object System", "ACM Transactions on Computer Systems", Vol. 16,
No. 1, February 1998 it was claimed that implementations of the Orca
language for parallel processing with distributed shared memory with
little reliance on operating systems was successful. Their Orca
implementations supposedly performed better than implementations of
other distributed shared models with libraries. However, unrelated to
whether a library or language approach was used, Orca has a very different
cache-coherence protocol which is much more relevant to relative
successes.

Ada is successful and the message passing "years ago with Ada" is
prohibited in for example Reliable Ada Verifiable Executive Needed for
Scheduling Critical Applications in Real-time which still permits
multitasking. Or do you contend that message passing is good if performed
by an operating system instead of a language?