-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stewart Stremler wrote:
> begin  quoting Darren New as of Wed, Jun 13, 2007 at 06:27:58PM -0700:
>> Paul G. Allen wrote:
>>> NT dose not support true preemptive multi-tasking. 
>> I don't think you know what preemptive multi-tasking means.
> 
> Dem's fightin' words!
> 
> "Preemptive multitasking" means that you don't have to think about what
> other processes may or may not be doing.  No yields necessary, no
> cooperation.  Write your programs to be right greedy little bastards,
> and the OS will make sure everyone gets to run appropriately.

No, it really doesn't. Preemptive multitasking just means that the OS
kernel may choose to context switch without you getting a thumbs up from
the current task. Yes, no yields are necessary or no cooperation, but on
the flip side there is no guarantee about when you will get a shot at
running next. It often means greater latency.

> Continuations and throwing in lots of "yields()" is a sign that one
> doesn't have preemptive multitasking in the target environment.

Hmm... The QNX guys will be very interested to learn this.

>> It doesn't mean "servicing interrupts promptly." That would be "real-time".
> Would it?
> 
> I thought real-time was all about determinism.  If it deterministically
> waited until next week, it ought to still be a candidate for
> "real-time", no? :)

Okay, so if you acknowledge a non-deterministic scheduling of tasks...
why are you surprised that tasks aren't scheduled as you determine they
should be?

- --Chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGcLIDOagjPOywMBARAmIpAJ4rw+QG4iC0O6dx8hW7eo+Tv5QBgACgr9c0
2Dbpe0UV5PZjgm/V4yPS6tk=
=T703
-----END PGP SIGNATURE-----

-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg

Reply via email to