Michael Abbott

Related thread: https://discourse.julialang.org/t/map-performance-with-cuarrays/33497/10 Are there functions for which some variant of `f.(eachcol(cu(x)))` does make sense? i.e. for which mapping over slices of a CuArray is useful & fast?

BTW, `M + I` is partially fixed by writing `Base.axes(x::Symmetric) = axes(parent(x))`, as this causes it to make an MMatrix to write into. Probably that should be done in any...

The base PR doesn't do anything for special matrices, since they are all square. But the cost of checking their sizes unnecessarily is usually negligible -- only tiny SMatrices are...

I was also surprised it didn't get compiled away. Maybe it just needs more things inlined, if I try `@eval LinearAlgebra @inline function _quad_matmul(A,B,C,D) ...` then the time difference in...

The `@inline` seems safe, but besides solving the above benchmark, I wonder whether it would solve real-world equivalents? Or whether `*` not inlining would reproduce the problem? > Just eliminating...

Xref #906 which has some discussion of what to print.

Hah I was going to have a go at this too, https://github.com/JuliaArrays/FillArrays.jl/pull/144 is the other one... But I think they can be more compact, at least sometimes. Why not print...

Here's a quick attempt to hook into the machinery... `@less Base.typeinfo_prefix(stdout, [1,2,3])` is where it decides what prefix to print on other arrays, thus: ```julia julia> function Base.typeinfo_prefix(io::IO, X::SArray) typeinfo...

Much as I like `SA`, it is a bit of a syntax trick. You could still minimise down to `SVector(1.0, 2.0, 3.0)` when the eltype is obvious. But `SMatrix{2,2}(1,2,3,4)` is...

Would you ever want an abstract type though? `Any[1]` you can write `2.5` into, maybe you could want that for an MArray? `SVector{2}(Any[1,"β"])` does preserve it, of course.