👍

We do the our own norm computation for a long fixed issues in numpy < 1.5. Our minimum numpy is 1.6 (and has been since mpl1.4) so we can remove all of that code + update the kwarg.

This is in principle easy, but touching hist can get surprisingly complex surprisingly quickly, so tagging as 'medium' for that reason. However, this change will be relatively self-contained (and should not require touching code outside of hist, hence 'new contributor friendly'.

There is a bunch of special casing farther down the function for when the histogram is normed that will need to be checked to still work.

I think we should differ from numpy a bit here and:

• set both normed and density to None by default.
• if both are not None raise
• covert to 'density' fully internally
• maintain default behavior of counts (not density)

Bit of git forensics

doing the norm our selves introduced in 2840392

The code around it has been refactored a bit and the note re-formatted, but it looks like it has not been fundamentally changed.

attn @efiring @neggert