#: 7257 S3/Languages
    09-Oct-90  09:45:20
Sb: #7250-#real time languages
Fm: Mark Wuest 74030,332
To: Pete Lyall 76703,4230 (X)

WHAT!? OS-9 not real time!? (Lucky for you none of their sales reps are on here
)
 Seriously, the thing that makes OS-9 (and VRTX, and PSOS, etc) real-time is
the predictability of performance when it's needed. That is the very thing HP
put into HP-UX to make a "real-time" version of Unix. If you always know
exactly (within reason) how long a process or code segment is going to take,
then it is real-time. OS-9 user-state processes don't qualify, as they are
subject to scheduling (time-slicing) and interrupts.
 What we have done is to put timing-critical processes in system state and let
them "share" by tsleep()ing if and when it knows it has nothing to do. Although
one could argue that we aren't using OS-9 (or any OS) while we are in system
state, (in fact, there are points where the system clock is masked) it is nice
to have the kernel there to facilitate things such as fork()ing the process in
the first place and providing a systematic interface to the outside world
(i/o).
 Geez, I can't believe I jumped into a discussion about "real-time" .
 Mark

There is 1 Reply.