TST: improve coverage of array.jl

[catawbasam]

No description provided.

[catawbasam]

For type-related tests, I think you want someone more comfortable with the logic of the type system (see #10255).
Shall I close this PR?

[nalimilan]

No, don't worry, you'll probably get this right, and people can help you if you hesitate. Also, tests are mostly meant to catch regressions: you can start assuming that the types of the current outputs are correct.

[catawbasam]

Ok, I'm hesitating and would appreciate some advice.

'===' uses pointer equality for mutable objects like arrays and strings. Given that, these results make sense:

julia> ca = complex([0,0,0],0)
3-element Array{Complex{Int64},1}:
 0+0im
 0+0im
 0+0im

julia> ca2=Complex{Int64}[0+0im, 0+0im, 0+0im]
3-element Array{Complex{Int64},1}:
 0+0im
 0+0im
 0+0im

julia> ca == ca2
true
julia> ca === ca2
false

But what is the right type check here? typeof()?

julia> ca==ca2 && typeof(ca)==typeof(ca2)
true

`

[jiahao]

That was what I had in mind.

[jakebolewski]

I would do all(map(is, ca, ca2)).

[catawbasam]

So

julia> ca3 = Complex[0+0im, 0+0im, 0+0im]
3-element Array{Complex{T<:Real},1}:
 0+0im
 0+0im
 0+0im

julia> all(map(is, ca, ca3))
true

Implying this test would pass even if the return type of complex([0,0,0],0) changed to Array{Complex{T<:Real},1} at some point.

Or do we need ca==ca2 && typeof(ca)==typeof(ca2) && all(map(is, ca, ca2))?

EDIT: Maybe better all(map(is, ca, ca2)) && typeof(ca)==typeof(ca2)

[catawbasam]

Is there an existing function that does the equivalent of all(map(is, ca, ca2)) && typeof(ca)==typeof(ca2)?
If not, maybe there should be.

[vtjnash]

i would check the arrayops.jl tests, i know there's usage of @test_approx_eq with a eps=0