Op 28-06-11 11:23, Lukáš Petrovický schreef:
I'd vote for 1) - when the solution is not fully initialized, it doesn't
fulfill my requirements for a solution. So, instead of calling it the
best solution, let's make it null.
Make sense. Do note that the best solution
might still break hard
constraints (depending on allocated planning time and testdata difficulty),
which many also consider that it doesn't furfill their requirements for
a solution.
OTH, there might not even be a solution that doesn't break any hard
constraints.
--
With kind regards,
Geoffrey De Smet