[coyotos-dev] Thoughts on FCRBs and asynchrony
Christopher Nelson
paradox at BBHC.ORG
Wed Feb 8 14:36:32 EST 2006
> On Tue, 2006-02-07 at 15:20 -0700, Christopher Nelson wrote:
>
> > My first question is, how do activations (or continuations)
> simplify a
> > program versus threads, when you consider that an activation (or
> > continuation) is basically just a special thread? In other
> words, if
> > you do anything that could interfere with the operation of the
> > previously interrupted line of execution, you still have to worry
> > about synchronization.
>
> In some uses they do not simplify. It depends on your
> programming idiom.
>
> One idiom where things DO simplify is the event-driven
> program. I can have lots of outstanding FCRBs. When the
> corresponding events arrive, I will enqueue them for later processing.
>
> In this idiom, you will discover that there are exactly two user-mode
> threads: the main worker thread and the thread that performs
> the enqueueing of events. The enqueueing thread won't require
> floating point state, etc.
I see. Thank you for helping me understand this.
-={C}=-
More information about the coyotos-dev
mailing list