Package ai.timefold.solver.core.api.solver.phase

Interface PhaseCommand<Solution_>

Type Parameters:

Solution_ - the solution type, the class with the PlanningSolution annotation

All Known Subinterfaces:

CustomPhaseCommand<Solution_>

@NullMarked
public interface PhaseCommand<Solution_>

Runs a custom algorithm as a Phase of the Solver that changes the planning variables. To change problem facts, use Solver.addProblemChange(ProblemChange) instead.

To add custom properties, configure custom properties and add public setters for them.

Method Summary

Modifier and Type	Method	Description
void	changeWorkingSolution(ScoreDirector<Solution_> scoreDirector, BooleanSupplier isPhaseTerminated)	Changes working solution of ScoreDirector.getWorkingSolution().

Method Details

changeWorkingSolution

void changeWorkingSolution(ScoreDirector<Solution_> scoreDirector,
 BooleanSupplier isPhaseTerminated)

Changes working solution of ScoreDirector.getWorkingSolution(). When the working solution is modified, the ScoreDirector must be correctly notified (through ScoreDirector.beforeVariableChanged(Object, String) and ScoreDirector.afterVariableChanged(Object, String)), otherwise calculated Scores will be corrupted.

Don't forget to call ScoreDirector.triggerVariableListeners() after each set of changes (especially before every InnerScoreDirector.calculateScore() call) to ensure all shadow variables are updated.

Parameters:

scoreDirector - the ScoreDirector that needs to get notified of the changes.

isPhaseTerminated - long-running command implementations should check this periodically and terminate early if it returns true. Otherwise the terminations configured by the user will have no effect, as the solver can only terminate itself when a command has ended.