Mathieu Besançon

@non-Jedi thanks for this issue. - [ ] As mentioned, https://github.com/JuliaStats/StatsBase.jl/issues/493 is a big one. Once done, we have to migrate all StatsBase functions - [ ] #823 made us...

@datnamer > What's needed for GPU support? I am not sure what you mean. In doubt, I'd say generic containers everywhere instead of hardcoded Float64

Refactoring tests is a major point blocking #823. I was considering 1.0 not only from the semver point of view, but also as signal of readiness and robustness of the...

> Not really: we could merge #823, it's just that the keyword constructors aren't really being tested. Not testing the keyword constructor would be problematic, we would merge something we...

@non-Jedi if you want to pick up some issues from there feel free :) I tagged a bunch of issues as up-for-grabs

Also: @cscherrer, this one is still open and could potentially lead to breaks https://github.com/JuliaStats/Distributions.jl/issues/771

It's not major, it could lead to breaking change but nothing awful, and always fixable by specifying types

Not if you add a deprecation fallback, but anyway yes this will be better done before 1.0 to be able to break it

@xukai92 would this imply a large sparse covariance matrix of the stacked random variables?

@nickrobinson251 thanks for keeping up on this. I tend to agree with your point, the two PR mentioned have been stalled and I'd be fine with keeping them for a...