Simeon David Schaub

Could we also add a two-arg version with a function argument? (Could just be implemented as `stack(Iterators.map(f, x))`.) I find myself using the `mapreduce(f, vcat, x)` pattern quite often, but...

Note that this isn't specific to tuples or arrays, but is the case for any kind of argument list: ```julia julia> x=1; println(x, x+=1) 22 ``` The underlying issue here...

> Seems like not too big of a deal to add a third one: > > ``` > -> ① + 1 > ``` I have to agree with @stevengj...

I would be curious to know whether #82 helps here. I am not expecting a huge difference, but it might help a little.

Since the leading dimensions here are quite small, #82 should improve this a bit, since that enables threading over more than just one dimension. Will probably still be nowhere near...

We should now be able to check the complex case with FiniteDifferences. But right now, most of our LinearAlgebra rules are restricted to real numbers anyways, so we could also...

Probably related to #46. Although we probably don't need the size information here.

Zygote doesn't use ChainRules for handling broadcasting at all, everything is defined here: https://github.com/FluxML/Zygote.jl/blob/master/src/lib/broadcast.jl. We'd first need to solve https://github.com/JuliaDiff/ChainRulesCore.jl/issues/68 before we can define such rules in ChainRules.

See https://github.com/JuliaDiff/ChainRulesCore.jl/issues/68, we currently lack a proper way for rrules to call back into the AD system, so ChainRules can't really deal with higher-order functions yet.

No, this is unrelated. (The reason `range` is in there is that I first thought ranges had to be quoted to, but that's not the case. Might test if their...