> > wonders how such an implementation can be validated for compliance with
> > annex D.
>
> Since the semantics are the same whether the same-priority tasks are
> running completely in parallel (such as on a multiprocessor) or not, I can
> not see any problem that could possibly be a result of time slicing.  Could
> you send some more information?

He didn't say "same-priority."  Unless the implementation "does the right
thing" the semantics are NOT the same and the implementation should not be
validated.

If you can not turn off O.S. time slicing, and if the O.S. time slicing
does not satisfy the Ada rules, then the Ada RTS must implement all its
tasks internally in one process.  That way, when the O.S. time-slicing
switches to another process, it is turning off ALL of your Ada tasks.
Then when it switches back to your program, the task that was running when
it switched out is still running.

As for a problem from time slicing:  I have seen environments where EVERY
process is guaranteed CPU time, but the lower priority ones just get less
of it.  Under the right circumstances, this is allocating resources to a
process that has a lower priority.  If "process" = "task" then this
violates D.1(15).