[gecode-users] Propagation for reasoning with models

[Guido Tack]

Hi,

the only solution I can think of is to postpone the actual posting of the constraints.  Create a data type that can represent the constraints you want to post, then collect the constraints that you want to postpone in a vector, and only when you need the propagation, you actually post them.  When you clone a space, copy the pending descriptions over to the clone.

Cheers,
	Guido

Nacho wrote:

> Hi all,
> We are using Gecode as a external FD solver for our system, which basically
> reason about models. Our system is implemented in Prolog, and we rely in the
> Prolog backtracking mechanism. We need to add and remove constraints from a
> Gecode Space. As it is not possible to remove constraints from a Space, you
> recommended me to clone each Space before any new constraint is added. 
> 
> Now we are planning a non-propagation mode for our system, in order to allow the
> user the possibility of propagate or not the different constraints which are
> added to the model. With this, we are trying to stablish a new configuration
> reducing the spent time in the solver interface layer. However, I can not
> inmediately implement this feature in my Gecode integration, as the Space
> becomes non-stable and this forbids the clonning process. Would it be possible
> to implement this feature in the context we are using Gecode for?  
> 
> Thank you very much in advance,
> Nacho
> 
> 
> _______________________________________________
> Gecode users mailing list
> users at gecode.org
> https://www.gecode.org/mailman/listinfo/gecode-users

-- 
Guido Tack, http://people.cs.kuleuven.be/~guido.tack/