I am working on a java 2d game. Keeping things short, I'd like to improve my current input handling. Basically the player can move up/down/left/right with WASD, perform an attack with mouse left click and shoot with mouse right click.
Following this article I implemented my own finite state machine like:
interface State {
State transition(Input input, double deltatime)
void update(Input input, Entity e, double deltatime)
}
where Entity is player (I'm using some variant of entity-component-system, but its not relevant here). Then a state would look like this:
class WalkState implements State {
State transition(Input input, double deltatime) {
if (input.isShooting())
return new ShootState()
if (input.isAttacking())
return new AttackState()
if (!input.isMoving())
return new IdleState()
return null;
}
void update(Input input, Entity e, double deltatime) {
if (input.isUp())
//move up
if (input.isDown())
//move down
...
}
}
The code that updates the player every tick looks like this:
if player is blocked executing animation
return; // do not react to new input
var currState = player.getState()
newState = currState.transition(input, deltatime);
if (newState != null)
player.setState(newState)
player.getState.update(input, player, deltatime)
// update animation by getting a string description from the state
My doubts are that maybe using a finite state machine for this is not optimal and that the input taking class reimains made of boolean fields to be checked. I know of command pattern but I thought states gave me more flexibility. For instance I could insert timers into some state to trigger extra actions. Any suggestion is appreciated
