Axis ticks misaligned when using integersOnly flag

[nickofthyme]

Describe the issue

Using the integersOnly flag with a small domain (i.e. 0 to 0.7) the axis ticks become misaligned with the scaled data.

Notice the value of the left-most bar is 0.68574 but the top tick is displayed as 1 when the true value is 0.7.

Expected behaviour
Ticks are accurately displayed to align with the axis scale.

Additional context
This seems to be a bug in the visible_ticks.ts logic. When I disable the adaptiveTickCount it works as expected. Do you know what this adaptive logic is doing and why it's computed independently from the actual scale?

elastic-charts/packages/charts/src/chart_types/xy_chart/state/selectors/visible_ticks.ts

Lines 283 to 327 in 3b3aa03

	const fillLayer = (maxTickCountForLayer: number) => {
	let fallbackAskedTickCount = 2;
	let fallbackReceivedTickCount = Infinity;
	if (adaptiveTickCount) {
	let previousActualTickCount = NaN;
	for (let triedTickCount = maxTickCountForLayer; triedTickCount >= 1; triedTickCount--) {
	const scale = getScale(triedTickCount);
	const actualTickCount = scale?.ticks().length ?? 0;
	if (!scale || actualTickCount === previousActualTickCount || actualTickCount < 2) continue;
	const raster = getMeasuredTicks(scale, scale.ticks(), undefined, 0, userProvidedLabelFormatter);
	const nonZeroLengthTicks = raster.ticks.filter((tick) => tick.label.length > 0);
	const uniqueLabels = new Set(raster.ticks.map((tick) => tick.label));
	const areLabelsUnique = raster.ticks.length === uniqueLabels.size;
	const areAdjacentTimeLabelsUnique =
	scale.type === ScaleType.Time &&
	!axisSpec.showDuplicatedTicks &&
	(areLabelsUnique || raster.ticks.every((d, i, a) => i === 0 || d.label !== a[i - 1]?.label));
	const atLeastTwoTicks = uniqueLabels.size >= 2;
	const allTicksFit = !uniqueLabels.has('');
	const compliant =
	axisSpec &&
	(scale.type === ScaleType.Time || atLeastTwoTicks) &&
	(scale.type === ScaleType.Log || allTicksFit) &&
	((scale.type === ScaleType.Time && (axisSpec.showDuplicatedTicks || areAdjacentTimeLabelsUnique)) ||
	(scale.type === ScaleType.Log
	? new Set(nonZeroLengthTicks.map((tick) => tick.label)).size === nonZeroLengthTicks.length
	: areLabelsUnique));
	previousActualTickCount = actualTickCount;
	if (raster && compliant) {
	return {
	entry: {
	...raster,
	ticks: scale.type === ScaleType.Log ? raster.ticks : nonZeroLengthTicks,
	},
	fallbackAskedTickCount,
	};
	} else if (atLeastTwoTicks && uniqueLabels.size <= fallbackReceivedTickCount) {
	// let's remember the smallest triedTickCount that yielded two distinct ticks
	fallbackReceivedTickCount = uniqueLabels.size;
	fallbackAskedTickCount = triedTickCount;
	}
	}
	}
	return { fallbackAskedTickCount };
	};

So taking your example, like below, you can see the issue. But if I log the domain, range and ticks for all iterations of the y axis. You can see the first 3 are determined using the correct scale, the last 3 are from the code above, which appears to be walking the domain out from 0.7 to 0.8 to 1. This returns the ticks but these ticks are used with the scale computed originally with a different domain. 🤔 What the??

data: (5) [
  ['Sensor 1', 0.68574]
  ['Sensor 2', 0.5857399999999999]
  ['Sensor 3', 0.48573999999999995]
  ['Sensor 4', 0.38573999999999997]
  ['Sensor 5', 0.28574]
]

Tick iterations of y axis continuous scale
-------------------------------------------
 domain: (2) [0, 0.7]
 range: (2) [0, 1]
 ticks: (8) [0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7]
-------------------------------------------
 domain: (2) [0, 0.7]
 range: (2) [307, 0]
 ticks: (8) [0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7]
-------------------------------------------
 domain: (2) [0, 0.7]
 range: (2) [307, 0]
 ticks: (8) [0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7]
-------------------------------------------
 domain: (2) [0, 0.8]
 range: (2) [307, 0]
 ticks: (5) [0, 0.2, 0.4, 0.6, 0.8]
-------------------------------------------
 domain: (2) [0, 0.8]
 range: (2) [307, 0]
 ticks: (5) [0, 0.2, 0.4, 0.6, 0.8]
-------------------------------------------
 domain: (2) [0, 1]
 range: (2) [307, 0]
 ticks: (3) [0, 0.5, 1]