Lilith Orion Hafner

This could be because I am misusing the `expand_derivatives` function. I'm happy to add a docstring if someone tells me what it it's supposed to do.

Bumping because now that Julia 1.9 is out, this is a pretty clear performance win in addition to a maintainability win.

I read that search result as 10 distinct packages using unnecessary parentheses in this syntactic construction, which is definitely not nobody, so I guess we have to keep support for...

@Keno, are GraphVis.jl or GraphVis_jll appropriate here?

Here are some benchmarks that informed my commentary: A value V > 1 indicates that :dict is V times faster than :radixsort. A value -V > 1 indicates that :radixsort...

it used to be a 100x performance loss to use radixsort on small sizes, now it's only about 1.5x, not worth the added complexity and runtime variability imo.

This bug is no longer present. ```julia julia> using StatsBase julia> n = 5030 5030 julia> N = 141230 141230 julia> sample(1:n, StatsBase.Weights(rand(Float32,n)), N, replace=true) 141230-element Vector{Int64}: 247 862 2144...

This optimization would be unsafe because it is possible for `make_alias_table!` to produce invalid alias tables: ```julia julia> StatsBase.alias_sample!(Random.default_rng(), rand(10), weights(fill(0, 10)), rand(10)) ERROR: BoundsError: attempt to access 10-element Vector{Float64}...

This issue is fixed on Julia master. It is still present on 1.9.0-rc2 because https://github.com/JuliaSparse/SparseArrays.jl/pull/296 hasn't landed there but should be fixed on 1.9.0. ```julia julia> using SparseArrays, ForwardDiff julia>...

(fixed on Julia 1.9.0-rc3)