Hi Liviu,

On Wed, Sep 16, 2009 at 2:23 AM, Ciortea Liviu <liviu.ciortea at epfl.ch> wrote:
> I noticed that the internal forks are not considered in the seeding scheme.

I'm not sure what you mean by this. When a state forks, the set of
seeds that state is following is split, regardless of whether the fork
is internal or not.

> This means, among other things, that there is not a one-to-one mapping 
> between a seed and a single state in the execution tree.

This is always true, when execution starts there all the seeds are
associated with a single state. As that state forks it the set of
seeds it is associated with shrinks.

> I am implementing a similar replay mechanism, based on seeding. However, in 
> my case I need to make sure I include the internal forks as well. Is there 
> any side-effect of removing the condition for the fork to be internal when 
> seeding? Also, since in internal forks the solver is called, would this be 
> deterministic?

I don't understand this. Are you perhaps talking about the replayPath
functionality, not the seeding functionality?

 - Daniel

Reply via email to