Re: FD Solver - incorrect behaviour

Hi Stephen,

you are right this kind of limitation occurs easily with non-linear 
constraints. It is difficult to find a solution which works in every 
case. A simple workaround is to slightly modify your source as follows:

bug(L) :-
	L=[P1,P2,P3,P4,P6,P7,P8,P9],
	fd_domain(L,[2,3,5,7]),
%	uncomment the following line and this works correctly
%	P2#=7,
	Z #= 100 * P3 + 10 * P2 + P1,
	P4 * Z #= 1000 * P9 + 100 * P8 + 10 * P7 + P6.

I try to fix things asap :-)


Stephen Meyles wrote:

> The following seems to be a bug with the FD solver:
> 
> bug(L) :-
> 	L=[P1,P2,P3,P4,P6,P7,P8,P9],
> 	fd_domain(L,[2,3,5,7]),
> %	uncomment the following line and this works correctly
> %	P2#=7,
> 	P4 * (100 * P3 + 10 * P2 + P1 )
> 		#= 1000 * P9 + 100 * P8 + 10 * P7 + P6.
> 
> With P2 constrained to [2,3,5,7], the predicate returns no.  With it
> further constrained to 7 alone, it correctly returns the variable list
> suitably constrained.
> 
> RH9 Linux (x86, glibc 2.3)
> gprolog v1.2.16 (stable) and v1.2.18 (unstable)
> 
> Regards
> 
> Stephen