[Stackless] Algorithm for Stackless Scheduler

Andrew Francis andrewfr_ice at yahoo.com
Fri Nov 21 23:31:30 CET 2008


Hi Tim:


--- On Fri, 11/21/08, Tim Kientzle <tim at metaweb.com> wrote:

> Compared to a typical OS scheduler, stackless scheduling is
> extremely lightweight, 

No argument here. My comment about context switching is relative 
to Stackless operations. Still fast.

> In my (somewhat limited) experience with Stackless,
> workflow issues dominate here.  There are some workflow
> patterns where it's very important that a sender be able
> to queue as many outgoing messages as possible before it
> relinquishes the CPU, others where it's more important
> that the receiver begin processing as quickly as possible. 

Agreed. It is these workflow patterns that I am trying to 
understand more in depth. 

In the case of the sender situation, (ll other things being 
equal) I believe the 'queuing' is desired if there is a relatively 
high overhead/fixed cost to executing the sender. However in this scenario 
one should take into consideration stuff like average response time and/or 
stability - is the system is sending output at least at the same rate 
that it is receiving input.

As a side note, it is interesting to figure out what the exact
difference between a queue and a channel with sender preference. 
Is a queue object really necessary?

> The "receiver preference" and "sender
> preference" knobs seem to provide a pretty good way to
> model those requirements.

Once again agreed. However preferences aren't really addressed in the Stackless documentation.

Cheers,
Andrew


      




More information about the Stackless mailing list