repeat(A::AbstractArray; inner=k::Int) fails for Arrays of Arrays.

[braydenware]

#20635 changed the implementation of repeat for arrays to improve performance, but it now results in errors for Arrays of Arrays.

On julia version 0.6.0,

julia> repeat([[1,2], [3,4]]; inner=2)
ERROR: MethodError: Cannot `convert` an object of type Int64 to an object of type Array{Int64,1}
This may have arisen from a call to the constructor Array{Int64,1}(...),
since type constructors fall back to convert methods.
Stacktrace:
 [1] setindex!(::Array{Array{Int64,1},1}, ::Array{Int64,1}, ::UnitRange{Int64}) at ./array.jl:573
 [2] _repeat at ./abstractarraymath.jl:411 [inlined]
 [3] #repeat#117(::Int64, ::Tuple{Int64}, ::Function, ::Array{Array{Int64,1},1}) at ./abstractarraymath.jl:366
 [4] (::Base.#kw##repeat)(::Array{Any,1}, ::Base.#repeat, ::Array{Array{Int64,1},1}) at ./<missing>:0

The result should be

julia> repeat([[1,2], [3,4]]; inner=2)
4-element Array{Array{Int64,1},1}:
 [1,2]
 [1,2]
 [3,4]
 [3,4]

as it is in julia 0.5.

The error message is different for higher dimensional Arrays:

julia> repeat([fill(1, 2,2), fill(2, 2,2)]; inner=2)
ERROR: DimensionMismatch("tried to assign 4 elements to 2 destinations")
Stacktrace:
 [1] throw_setindex_mismatch(::Array{Int64,2}, ::Tuple{Int64}) at ./indices.jl:92
 [2] setindex_shape_check(::Array{Int64,2}, ::Int64) at ./indices.jl:140
 [3] setindex!(::Array{Array{Int64,2},1}, ::Array{Int64,2}, ::UnitRange{Int64}) at ./array.jl:563
 [4] _repeat at ./abstractarraymath.jl:411 [inlined]
 [5] #repeat#117(::Int64, ::Tuple{Int64}, ::Function, ::Array{Array{Int64,2},1}) at ./abstractarraymath.jl:366
 [6] (::Base.#kw##repeat)(::Array{Any,1}, ::Base.#repeat, ::Array{Array{Int64,2},1}) at ./<missing>:0

which should be

julia> repeat([fill(1, 2,2), fill(2, 2,2)]; inner=2)
4-element Array{Array{Int64,2},1}:
 [1 1; 1 1]
 [1 1; 1 1]
 [2 2; 2 2]
 [2 2; 2 2]

The culprit seems to be trying to set a slice of the output array at once, using the behavior A[vector] = scalar to set multiple elements of A to the same scalar.
When A is an Array of Arrays and scalar is an Array, A[vector] = scalar behavior does not work, perhaps because it would be ambiguous with A[vector] = vector.
For example,

R = Vector{Matrix{Int}}(4)
R[1] = fill(1, 2,2)
R[2] = fill(1, 2,2)
R[3] = fill(2, 2,2)
R[4] = fill(2, 2,2)

cannot be replaced with

R = Vector{Matrix{Int}}(4)
R[1:2] = fill(1, 2,2)
R[3:4] = fill(2, 2,2)

for producing the behavior of repeat([fill(1, 2,2), fill(2, 2,2)]; inner=2).

Note that the behavior of the outer keyword works fine:

julia> repeat([[1,2],[3,4]]; outer=2)
4-element Array{Array{Int64,1},1}:
 [1, 2]
 [3, 4]
 [1, 2]
 [3, 4]

as well as the behavior of indexable objects that are not AbstractArrays:

julia> repeat([(1,2),(3,4)]; inner=2)
4-element Array{Tuple{Int64,Int64},1}:
 (1, 2)
 (1, 2)
 (3, 4)
 (3, 4)

[braydenware]

Oops, didn't see that you filed an issue slightly before me. Should I close?

[pabloferz]

No, I sent a PR to fix the issue, this will be closed once the PR gets merged.