#fork and deterministic resumption of the resulting process
Yoshiki Ohshima
yoshiki at vpri.org
Fri Feb 8 06:50:25 UTC 2008
> I'd like to emphasize this (please, increase the font size, when you
> reading following phrase):
>
> ... a program that relies on a particular implementation of
> scheduling is wrong.
There is still some mismatch.
Sure, at one level, you can say that if you want to ensure some
ordering in a concurrent program, use a proper concurrency control
mechanism; don't rely on the implementation detail.
But at another level, you can say that if you write your own
scheduler, use the full-knowledge of it. There scheduler is just
another module of your system. If using that knowledge makes the
product rock-solid, there is nothing wrong with it.
And, with Andreas patch, normal programmer doesn't have to assume
the scheduling ordering. His patch doesn't prevent people from using
a proper concurrency mechanism.
In general, I agree that your statement above is a good principle,
but in current Squeak context, there is another path, too.
-- Yoshiki
More information about the Squeak-dev
mailing list
|