[Stackless] Santa concurrency problem

[Christian Tismer]

Hi Carlos,

sorry for answering late.
I'm slightly under water these days.

On 02.04.2007, at 20:43, Carlos Eduardo de Paula wrote:

> Returning to our santa example, I tried to use the new Timing  
> approach on it.
>
> I applied it monkeypatching the sleep function but found a problem,
> when using the old approach (ManageSleep) the app runs forever but
> when I started using this one, the app stops running after random
> time. I think its a problem related to the management of the blocked
> tasklets...
>
> The code can be seen here:
>
> http://dpaste.com/hold/7800/
>
>
> Carlos
>
> On 3/30/07, Christian Tismer <tismer at stackless.com> wrote:
>>
>> On 30.03.2007, at 18:37, Carlos Eduardo de Paula wrote:
>>
>> > Hi Chris,
>> >
>> > take a look at the changes I made in QueuedTimer, It now uses a
>> > heapq.. I kept the old way commented just for clarification.. Do  
>> you
>> > think it performs better now or its the same.
>>
>> looks pretty good.
>> I think we simply use it and trash my silly dict approach.

It is even possible that it is  slightly slower, but superior in the
asymptotic behavior. We could of course try to measure that,
but I doubt it makes sense.

ciao - chris


_______________________________________________
Stackless mailing list
Stackless at stackless.com
http://www.stackless.com/mailman/listinfo/stackless