Alexander Plavin

What's the decision, if any? Currently, `var` is assumed to be circular: ```julia julia> using Distribustions julia> d = VonMises(0, 2) julia> var(d) 0.30222534203599194 julia> var(rand(d, 10^6)) 0.7641079384821371 julia> using...

For convenience, I added `Circular.var()` etc overloads for this distribution to `DirectionalStatistics.jl` – to return circular variance and std. Now, we have [there](https://gitlab.com/aplavin/DirectionalStatistics.jl/-/blob/670cb441b6c107144d649368bd4752c599ea1efe/test/runtests.jl#L369-376): ```julia Circular.mean(d) == Circular.mean(rand(d, N)) Circular.var(d) ==...

bump...

Many interval operations are "easy enough" to do without a special interval type at all, by working with endpoints directly. The whole point of an interval type (same with many...

> most distributions have a fixed support that can't be adjusted based on a given interval Not sure if I understand this part. Of course it doesn't make sense to...

Yes, this kind of probability calculations is convenient sometimes, but orthogonal to this PR. Here, I add intervals support to already existing functions, no new functions added. For playing with...

Gentle bump... This PR adds support for intervals arguments to existing methods where it makes sense and is totally unambiguous. Everything else including @chelate's suggestions is out of scope here...

Oh, sorry, I forgot to explicitly state the motivation to put this into Distributions indeed! This extension is much more likely to require changes due to changes in Distributions, than...

In case others come looking for this functionality: the interval support (this PR) is available in https://github.com/JuliaAPlavin/DistributionsExtra.jl. Not sure if I want to register that package yet... It has more...

That's what I thought originally: regular reals measure for `Continous`, counting measure for `Discrete`. But what the measure is for `censored()` then?