TemporalActions don't always end on ending value

Due to floating point arithmetic, TemporalActions that have an end argument aren’t guaranteed to end on that state.

Arbitrary example (Kotlin):

val action = ColorAction()
action.color = Color(0xffffffff.toInt())
action.duration = 1f
action.interpolation = Interpolation.pow5Out
action.endColor = Color(0x444444ff)

After execution, I would expect the actor’s color to be 0x444444ff; however, it’s 0x434343ff.

Would it make sense override end and do the equivalent of color.set(endColor) for all affected actions?

[obigu]

Would it make sense override end and do the equivalent of color.set(endColor) for all affected actions?

Need to be careful as the action may be reversed in which case the final color on action complete is the start color. I think it’s safer to deal with this inside the update for the special case where percentage == 1f.

[obigu]

@NathanSweet My comment was more oriented to the fact that start and end Color should be either both Colors instances or both be stored as their float components more than about the necessity of exposing the getter for the start color (haven’t suggested having a setter). After some more thought, I think the argument for minimal memory allocation outweights the benefits of using Color instances and, imo, end Color should also be stored as its float components (maybe lazily created if getEndColor() is called)