Andy Dienes

I imagine this was changed by https://github.com/JuliaLang/julia/pull/59165, and since it's tagged `minor change` it also seems the difference in behavior is intentional. I did not really follow neither the motivation...

> eliminating both nondeterminism and the effect of redefining methods in the same session there are several open issues observing inference changes when methods are redefined; does this PR affect...

here is another good stressor ``` julia> r1 = LinRange(0.3376448676263234, 1.509664528429199, 3); julia> r2 = range(0.3376448676263234, step=0.5860098304014378, length=3); julia> isless(r1, r2) || isless(r2, r1) || isequal(r1, r2) false ``` I...

are these accounts all from the same entity ...?

perhaps the issue is here ``` _to_same_csc(::AbstractSparseMatrixCSC{Tv, Ti}, V::AbstractMatrix, I...) where {Tv,Ti} = convert(SparseMatrixCSC{Tv,Ti}, V) ``` that `V` could be `map(length, I)` like it is for the `AbstractVector` case. I'm...

I am not a heavy user, but I think https://github.com/Roger-luo/Moshi.jl has a pretty well-thought-out design for reference

to bikeshed the term `Except`; that choice reads to me a bit like "returns everything _except_ `Foo`" which is kind of confusing. maybe `Result` to match Rust? other random ides:...

> I'm thinking about using break x instead, then return is available for function return. I do think we would need the multi-break syntax as well if we do that...

> but would break pattern matching on pairs it wouldn't "break" so much as "be super confusing" I guess? since right-associativity of `=>` means the first one left-to-right is for...

I bisected the new warning to https://github.com/JuliaLang/julia/pull/49637 . this doesn't really say much about root cause but at least it might explain the difference between float64 and float32