[comp.realtime] POSIX realtime standard

map@honda.siemens.com (Michael Platoff) (01/03/91)

I remember reading about an IEEE POSIX standard for a real-time
OS interface.  Does anyone know the progress of that effort?  Has
anything been drafted yet?  Does anyone have an estimate of
when real-time execs will begin to have a standardized interface?

Since I don't read this newsgroup very often, e-mail is the best
way to reach me.  I'll summarize and post any responses, if requested.

Thanks,

Michael Platoff
Michael Platoff               email: map@cadillac.siemens.com
Siemens Corporate Research    phone: (609) 734-3354
755 College Road East
Princeton, NJ 08540-6668

bog@lynx.uucp (Bill O. Gallmeister) (01/04/91)

map@honda.siemens.com (Michael Platoff) writes:

>I remember reading about an IEEE POSIX standard for a real-time
>OS interface.  Does anyone know the progress of that effort?  Has
>anything been drafted yet?  Does anyone have an estimate of
>when real-time execs will begin to have a standardized interface?

Michael said to send mail, so I did.  But I also thought this would
be of general interest, so I'm posting it.

POSIX 1003.4 is the real-time standards group, and Posix 1003.4
is the proposed standard.  Draft 9 of this proposal went out to ballot
about a year ago, with chapters on:

	priority scheduling
	binary semaphores
	memory locking
	shared memory
	asynchronous event notification
	clocks and timers
	IPC message passing
	synchronized I/O
	asynchronous I/O
	realtime files

We are planning on wrapping up the ballot resolution in New Orleans
next week (1/7/90 - 1/11/90).  After that, the draft will be sent out
(draft 10, now) for recirculation.  And the whole process will begin
again.  Once we hit a certain threashold of acceptances (I don't remember
the figure), the draft will become a standard.  I'm anticipating another
six months at least before that time.

Another proposed standard that just went to first ballot is the threads
draft (IEEE POSIX 1003.4a draft 5).  POSIX threads provide multiple threads
of control within a single UNIX (actually POSIX) process.  The draft has
chapters for

	thread management (create, exit, join, etc)
	thread scheduling
	thread synchronization (currently mutexes and condition variables)
	thread cancellation
	thread-specific data
	process behavior in the presence of threads
	signals and asynchronous events
	reentrant functions

Threads provide hope for reasonable Ada compilers under UNIX.  They're
also a much better way of dealing with asynchrony than are signals,
select, and all that nonsense.  In my humble opinion.

Threads just went to ballot (sent off the week before Christmas)
and will be a while in getting out as a draft standard.

Meanwhile, at least one company is already putting out a system that
supplies the functions found in Draft 9.  That's us;  our release 2.0
will have every chapter of POSIX.4 covered.  We don't anticipate a big
deal in going from draft 9 to draft 10.  We are also providing all the
functions of Posix 1003.4a drafts 3 and 4.  And we have two Ada compilers
running using threads, Alsys and DDC-I.

OSF also recently made threads-like noises, and DEC anounced plans for
at least P1003.1 support under VMS (!).  As far as I know, we are the
only ones out with POSIX.4.

For copies of the draft, you should write to the IEEE:

IEEE Service Center
445 Hoes Lane
Piscataway, NJ 08855-1331
(800) 678-IEEE
(201) 981-0060

Regards,

- bog
-- 
Bill O. Gallmeister                                     Lynx Real-Time Systems
bog@lynx.com                                            Trendy Los Gatos, CA