d3.histogram shouldn’t coerce ordinal input to numbers.

[cpietsch]

Lets say you have a
let x = d3.scaleTime()

and you want to use it on an histogram

d3.histogram()
  .value(d=> d.date)
  .domain(x.domain())
  .thresholds(x.ticks(d3.timeYear, 1))

[mbostock]

Yes, that’s valid. Is there a question? Here’s an example:

[cpietsch]

Oh sorry. I got confused by the negative x0/x1 values. Was expecting a date object or a positive value.

[mbostock]

Ah. Well yes, the bin.x0 and bin.x1 are coerced to numbers. And so are the other parts of the histogram (the domain, the values, the thresholds). However, it occurs to me that this is overly strict, and really the histogram only requires comparable (ordinal) values and not numbers, so we could relax that and leave dates as dates.

[lgrkvst]

Observation: if .value() returns Date objects, the first element's x0 and the last element's x1 will stay Date objects. The bin thresholds in-between is coerced into numberobjects.

var A = [1476448344837, 1476448348221, 1476448350581, 1476448352482];
var yay = d3.histogram().value(a => a)(A); // all numbers
var nay = d3.histogram().value(a => new Date(a))(A); // Date objects for `nay[0].x0` and `nay[1].x1`