David Widmann

https://github.com/JuliaStats/Distributions.jl/pull/1144 was merged, so the next release of Distributions will allow to use `loglikelihood(d::Distribution, x)` consistently for both single and multiple samples `x`.

Hej! It's not exactly clear to me what you want to do. Do you want to compute the expectation E f(X) for some function f and random variable X, where...

> Although GalacticOptim has Flux and Optim as two rather large dependencies... Depending on your exact needs, you could only depend on SciMLBase but not on GalacticOptim. In SciMLBase we...

I assume this is caused by FileIO 1.6.0 which was a major rewrite of FileIO (but supposed to be non-breaking - maybe RDatasets uses some internals of FileIO?). Downgrading to...

BTW since I haven't investigated what the actual problem is I am not sure if it is a problem of RDatasets or rather of RData.

The error message that I get when running the `iris` example is similar to https://github.com/JuliaData/RData.jl/issues/57 - maybe it's related to the gzip problem mentioned therein? Maybe uncompressed datasets work fine...

They are all caused by HypothesisTests. The first one due to some simple type stability issues (checked locally that fixing them fixes the example) and the second one since the...

Yes. There was no `else` branch. I fixed it in the PR, it was the main reason for the `check_tail` function and replacing the last `elseif` with `else`.

> Have you tried only replacing yi^4 with yi^2 * yi^2? Maybe `(yi^2)^2` would be more performant? Or would it be more difficult for the compiler to optimize?

Bump 😃 What is the general opinion? Should this change be applied, even if FileIO already fixed the original issue?