As a user , when I set a selection, we display only the data in the selection and adapt the scale to zoom in data

If we feed the histogram with some buckets where the value is no-data, the histogram throws "NaN" errors

The interval between buckets in this example is 0.5 knot. The tick '3' is the result of rounding '2.5'

Inspired by : https://bl.ocks.org/mbostock/5537671

create a new component to show a scale with 0 and max value, and a
threshold and a current value .
Three data input:

• max value
• current value
• current threshold

Other input:
-style of ticks

For now this rule is applied if multiple nodes values are selected on the same ring level. This rule will be applied also if only one value is selected several times on the same ring

DonutArc interface is renamed to TreeNode
ringName attribute is renamed to fieldName
name attribute is renamed to fieldValue
The other attributes stay the same

If the current donut selection is on 2 or more arcs, when you select an arc without 'child' it's throw an exception :

And totally freeze the component :

• Add keysToColors and colorsSaturationWeight attributes in DonutParams
• Enhance ColorGenerator interface

duplicate of #128

it corresponds to upgrade the latest modules used in our library :

"d3-array": "^3.2.0",
"d3-time-format": "^4.1.0",
"d3-hierarchy": "^3.1.2",
"d3-scale": "^4.0.2",
"d3-shape": "^3.1.0",
"d3-axis": "^3.0.0",
"d3-brush": "^3.0.0",
 "d3-interpolate": "^3.0.1"

The aim of keeping them is to remember what has been selected but not plotted anymore because of a fitler

This attribute contains the metric value represented on the node: count, avg, max, sum...

Before this, the metric value was set in size attribute. This was correct as we represented only the count on the donut nodes. Now we can also represent the other metrics on a donut. For this, the size of children nodes can be normalized, to have a coherent plot. The actual metricValue is then set in metricValue

As sum of leaf nodes can be less than the parent nodes, the sum policy in hierarchy is not convenient. So each node has to take its own size. For this the data input must specify the size of each node and not only the leaf nodes

Add a property to TreeNode object : fieldLabel.
This property will be used in layout for powerbars and donuts tooltip.
The property fieldValue will be used to filter (like now).

The field label will be calculate from fieldValue throught a process system with eval js, like we do in resultlist for column and detail.
If no process set in tree conf, fiedLabel = fieldValue

• Labels are too close to the ticks when they are on the top of the histogram
• Labels are too far from the ticks when they are on the bottom of the histogram

Up to now, all selected intervals are taken into account in the X axis domain =>
When an interval is selected (by double click) and there is no data in this interval, it will be considered in x Axis

Add an option that allows to display only data and current selection brush when there is no data on the fixed intervals.
This option is relevant when there two (or more) histograms on the same field. For instance timeline/detailed-timeline OR timeline/temporal-swimlane