Concurrency problems/solutions (was Re: Multy-core CPUs)

Sebastian Sastre ssastre at seaswork.com
Fri Oct 26 17:36:44 UTC 2007


Jason it's a good proposal.

For my part today, with the help of a friend of mine, I think I've found a
conceptual problem on the model I was exploring so I have to (re)think if it
continues to be valid at all.

Cheers,

Sebastian
PS: it's not about it's overhead nor hardware resources.

> -----Mensaje original-----
> De: squeak-dev-bounces at lists.squeakfoundation.org 
> [mailto:squeak-dev-bounces at lists.squeakfoundation.org] En 
> nombre de Jason Johnson
> Enviado el: Viernes, 26 de Octubre de 2007 01:39
> Para: peter at smalltalk.org; The general-purpose Squeak developers list
> Asunto: Concurrency problems/solutions (was Re: Multy-core CPUs)
> 
> Hi all,
> 
> There has been a lot of discussion about concurrency, and 
> what tends to happen is we all have ideas in our head of what 
> we want solved and why a given solution can/can't solve it.
> 
> So what I propose is, we come up with concurrency problems 
> and then how we would solve them in the different 
> methodologies.  With code. :)
> 
> Of course it is important to be specific.  An example is the 
> "million node" problem highlighted by Peter.  I would need to 
> know what this data is, where it comes from and so on so I 
> can determine if any one node should *ever* have the graph at once.
> 
> This is another challenge, it will be tempting to make 
> assumptions about what can be done with the data.  Please 
> don't. :)  You may have some reason you think the data can't 
> be partitioned, but perhaps it can.
> 
> I'm curious to see how other solutions solve the various 
> problems out there, and before I go do a bunch of work on the 
> Actor model I would love to see if there are any cases it 
> really can't deal with.
> 
> Thanks,
> Jason
> 




More information about the Squeak-dev mailing list