branchvalues='total' too strict

[nicolaskruchten]

It seems to be a common problem when using sunburst and treemap from Python and R that the values sum up in the original language but not in JS after passing through JSON, with errors in the 1e-14 range, e.g. https://codepen.io/nicolaskruchten/pen/MWYYJba

How about we accept the data if the difference is, say, less than 0.01% of the parent?

[alexcjohnson]

Let's use 1ppm difference. We have a constant for this - only used one other place ATM but at least it's a precedent:

plotly.js/src/constants/numerical.js

Lines 47 to 50 in fb8b28a

	/*
	* Are two values nearly equal? Compare to 1PPM
	*/
	ALMOST_EQUAL: 1 - 1e-6,

[mwouts]

Sure, I prefer 1ppm over 1%, at least visually we cannot make the difference (and in my experience multiplying each child weight by 1+1e-15 when computing the parent weight appears to be enough to fix the issue, so 1ppm is probably more than enough).

Also I was wondering if

1. we could get an error message on the plot when this kind of problem arises (with a blank plot I was not sure what the problem was)
2. or, maybe it would be just ok to show a parent that is smaller than the sum of children? (i.e. the opposite of what we have at https://plot.ly/python/sunburst-charts/#basic-sunburst-plot)

[nicolaskruchten]

we could get an error message on the plot when this kind of problem arises (with a blank plot I was not sure what the problem was)

There is an error message in the javascript console already, but it's very challenging for us to build a system to get access to those messages from Python or R. See plotly/plotly.py#1967 for a possible technical approach.

or, maybe it would be just ok to show a parent that is smaller than the sum of children? (i.e. the opposite of what we have at https://plot.ly/python/sunburst-charts/#basic-sunburst-plot)

It's not clear what that would actually look like if the parent is visibly smaller than the children :)