David Widmann
David Widmann
I don't see a clear inconsistency, there's no restriction on the values and `Inf`/`NaN` propagate in the same way for weighted operations: ```julia julia> using StatsBase julia> mean([1,2,3,Inf]) Inf julia>...
The Documenter failure is known (see e.g. https://github.com/JuliaStats/StatsBase.jl/pull/863#issuecomment-1533449124) and fixed upstream.
Can you point me to where exactly you use something that's removed in this PR? As mentioned above, based on the Juliahub search results it seems these types are only...
> Not all code is indexed by JuliaHub or even public. True, but I guess JuliaHub is the best way to judge practical implications of the PR. And based on...
> "Technically non breaking" is not great for something near the root of the dependency tree. The types are neither exported nor documented, so IMO this PR is clearly non-breaking....
To me it seems the main reason for tagging a breaking release in that case (even though it was a non-breaking change) was that it was clear from the publicly...
It's still not clear to me why we should add even more code to the already lengthy implementation and spend the time and effort to try to do it in...
> This seems fragile as lots of irrelevant changes in Base can break these tests. The point is that in contrast to the current master and alternative proposals this PR...
I guess such generalizations (and possible renaming) could be useful, but my initial feeling is that such more significant changes should maybe be left for follow-up PRs?
So, what's the conclusion here? What's missing or to resolve?