[BUG] Stacked bar using time axis without having the same datetime

Expected Behavior

Create a stacked bar using time axis with different datetime in the datasets. For example:

datasets: 
  [
    {
      label: 'A',
      data: 
      [
        {
          x: "01-01-2018",
          y: 12
        },
        {
          x: "02-01-2018",
          y: 15
        }, 
      ],
    },
    {
      label: 'B',
      data: 
      [
        {
          x: "02-01-2018",
          y: 23
        },
        {
          x: "03-01-2018",
          y: 34
        },
        {
          x: "04-01-2018",
          y: 45
        },
      ]
    }
  ],

In this example dataset B haven’t the point at “01-01-2018”

Current Behavior

Currently, using the stacked bar with time axis need the datasets having the same point datatime even if some point is set to 0.

datasets: 
  [
    {
      label: 'A',
      data: 
      [
        {
          x: "01-01-2018",
          y: 12
        },
        {
          x: "02-01-2018",
          y: 15
        }, 
      ],
    },
    {
      label: 'B',
      data: 
      [
        {
          x: "01-01-2018",
          y: 0
        },
        {
          x: "02-01-2018",
          y: 23
        },
        {
          x: "03-01-2018",
          y: 34
        },
        {
          x: "04-01-2018",
          y: 45
        },
      ]
    }
  ],

Possible Solution

Calculate the height of the single bar looking for every datasets if the datetime is present or not

Steps to Reproduce (for bugs)

Here the problem on stackoverflow Here the jsfiddle

Context

I want show the stacked bar without having to manipulate all the datasets for add the missing point.

Environment

• Chart.js version: 2.7.2
• Browser name and version: Firefox 61.0.1

[colinscz]

+1

This is also bugging us in our project. It would be great if you guys could provide a fix for this. Otherwise we always have to fill up the datasets, which lack certain ticks and set there a value ‘0’, which is then unfortunately also shown in the unified tooltip as value 0 for that particular dataset.

We want to dynamically load a bunch of different records (last month, last week, etc.) these contain diverse datasets per chunk. These datasets are different in length/number. Therefore we have to determine the order and enhancing of the datasets ourselves and this would be necessary with every reload/view change.

As seen here:

Here is a stackblitz for it: Stackblitz

[zicons]

+1