UHI histogram kind

[0ctagon]

The ROOT pythonization of TH1 decided to set the kind of the histogram to COUNT if unweighted, and MEAN if weighted (see here in ROOT).
In uproot, the kind.MEAN is only used for TProfile histograms, otherwise it's always a kind.COUNT (see here in uproot).
I was wondering why is there a difference in treatment between ROOT and uproot? kind.MEAN has a different treatment of variance (see UHI doc), so I was wondering which definition was correct between the two package?

[henryiii]

COUNT is for a summation histogram, and MEAN is for a profile.