[Choco-users] (no subject)

[Malcolm R. <mal...@cs...>]

I'm trying to make a constraint based representation of a list of  
variable length. Basically, I have an array of, say, 100 element  
variables V_0, ..., V_99, plus a length variable L in [0,100].  
Semantically, the first L of the element variables are in the list,  
the others are irrelevant.

I then have a bunch of different constraints which apply to all  
elements in the list (and only those). So essentially, for each such  
constraint C(V), I form a list of 100 constraints which state whether  
the constraint applies to each element, depending on the length  
variable:

L > 0 => C(V_0)
L > 1 => C(V_1)
...
L > 99 => C(V_99)

It seems to me that introducing 100 constraints like this is  
inefficient. I'd much rather have a single constraint which watched  
as the lower bound of L increased and added individual element  
constraints on the fly. So when L is foung to be greater than 2, C 
(V_0), C(V_1) and C(V_2) are added to the problem.

The apparent difficulty with this is the backtracking behaviour. If  
on backtracking we determine that L is less than 2, then we need to  
retract C(V_2) from the constraint store. Is there any easy way to do  
this in Choco?

Malcolm

--
Many clever men like you have trusted to civilisation.
Many clever Babylonians, many clever Egyptians,
Many clever men at the end of Rome.
                               - G.K.Chesterton, The Napoleon of  
Notting Hill