Michael Abbott

Xref #45793 which added `insertdims`. And I requested review from a few of those who commented on that PR.

Pre-0.12 bump? After #56637 this would need to be split into two, but I won't bother if nobody will review it.

No, only half of this PR is LinearAlgebra, the other half is Base. It can stay open for discussion?

Error is this: ```julia MethodError: repeat(::SparseArrays.SparseMatrixCSC{Float64, Int64}, ::Int64) is ambiguous. Candidates: repeat(A::SparseArrays.AbstractSparseMatrixCSC, m) @ SparseArrays /cache/build/tester-amdci4-14/julialang/julia-master/julia-d4b18e6530/share/julia/stdlib/v1.12/SparseArrays/src/sparsematrix.jl:3974 repeat(A::AbstractArray, counts::Integer...) @ Base abstractarraymath.jl:435 Possible fix, define repeat(::SparseArrays.AbstractSparseMatrixCSC, ::Integer) ``` Unfortunately I imagine...

Instead of fighting packages over ambiguities, 3ee4ffc changes it to an explicit check with an error. (There's already a function called `check` performing other sanity checks, such as negative repeats.)...

Yes. One advantage is that this also checks keywords inner/outer. Those can't easily have tight signatures as they accept almost anything: ```julia julia> repeat([1 2 3]; inner=transpose(2:3)) 2×9 Matrix{Int64}: 1...

Xref #41853, about adding static `Zero` & `One`.

> array concatenation works for vcat but not for hcat I think the problem is `hvncat`, while `hcat` and `vcat` are both fine: ```julia julia> [A B] isa SMatrix true...

Curiously the error goes away for a matrix instead of vector input: ```julia julia> x = rand(Float32, 3, 1); # now a Matrix julia> function forward(clf::Dense, x)::Float32 # signature widened...

The code that `Flux.gradient(f, ::Duplicated)` ends up calling is here, `make_zero!` then `autodiff`: https://github.com/FluxML/Flux.jl/blob/0e36af98f6fc5b7f3c95fe819a02172cfaaaf777/ext/FluxEnzymeExt/FluxEnzymeExt.jl#L44-L51