Animation fails with duration:0

[pixelzoom]

If an Animation is created and run with duration: 0, it fails with NaN due to a divide-by-zero error.

This was the cause of https://github.com/phetsims/unit-rates/issues/215. In Unit Rates, there's an animation that moves a marker along the number line, based on the number that's entered into a keypad.

Here's the Animation definition in DoubleNumberLineAccordionBox:

markerEditorAnimation = new Animation( {
  duration: 0.002 * Math.abs( destinationX - markerEditorNode.x ), // 2ms per 1 unit of distance
  easing: Easing.QUADRATIC_IN_OUT,
  object: markerEditorNode,
  attribute: 'x',
  to: destinationX
} );

If a value that is larger than the number line's range is entered into the keypad, then the marker moves to the far right end of the number line. If you enter 2 such large values in a row, then the distance that the marker needs to move is zero, and the computed value of Animation option duration will also be zero.

For example, here's the marker located at "333", and I'm about to enter "334". This requires no movement of the marker, so duration is 0.

When you try to run an Animation with duration: 0, Animation step will have a divide-by-zero at line 312, where this.length is zero and ratio is therefore NaN.

312 const ratio = Utils.clamp( ( this.length - this.remainingAnimation ) / this.length, 0, 1 );

This causes failure further along in the code. In the case of Unit Rates, it failed down in Easing.polynomialEaseInOutValue, where an assertion noticed that the ratio was invalid.

In Unit Rates, I'm simply not going to create an Animation if duration is zero. But this should be handled in Animation by either:

(a) an immediate assertion failure if options.duration is zero, or (b) the animation is a no-op and completes immediately when it's started.

I don't know which option is better, but option (b) would have prevented having to add logic to Unit Rates.

[jonathanolson]

I'll lean towards (b) unless it seems unworkable in the code. Thanks for spotting!

[jonathanolson]

Implemented (b) above.

[pixelzoom]

I tested this by removing the workaround in phetsims/unit-rates#215. Seems to work great. Thanks, closing.