Any[1 [1,2]] ERROR: DimensionMismatch

[egoecho]

In Julia 0.4.2,

julia> Any[1 [1,2]]
ERROR: DimensionMismatch("mismatch in dimension 1 (expected 1 got 2)")
 in cat_t at abstractarray.jl:850
 in typed_hcat at abstractarray.jl:878

The old style is fine:

julia> {1 [1,2]}
WARNING: deprecated syntax "{a b ...}".
Use "Any[a b ...]" instead.
1x2 Array{Any,2}:
 1  [1,2]

I also found that Any[1 [1]] = {1 1} rather than {1 [1]}

julia> Any[1 [1]]
1x2 Array{Any,2}:
 1  1

julia> {1 [1]}

WARNING: deprecated syntax "{a b ...}".
Use "Any[a b ...]" instead.
1x2 Array{Any,2}:
 1  [1]

[anthonyclays]

Any[1 [1, 2]] will call typed_hcat which makes julia attempt to make one large array composed of the smaller ones. The same thing will happen when you do simply [1 [1, 2]], or [1; [1 2]].

There's not really an exact equivalent of {...}. I believe your best bet would be to use reshape(Any[1, [1, 2]], 1, 2).

[KristofferC]

Imo, if you specify "Any" then there shouldn't be any concatenation.

[anthonyclays]

@KristofferC You could specialize the typed concatenation functions when typeC = Type{Any}, but then again what should the behaviour of AbstractArray[[1] [1, 2]] be?
You can't simply special-case Any (like {...} used to do). It's not really obvious what the best behaviour would be here.

[anthonyclays]

There will be ambiguity when using any subtype of AbstractArray in the construction of an Any-typed array, I believe (for example Any[1:3 10] fails too)

[jdlangs]

This looks like more evidence there should be parallel syntaxes for array construction and concatenation. (#7128)

[tkelman]

Or concatenation should be a function call rather than syntax. We should also flip the switch on the concatenation deprecation behavior from #8599

[JeffBezanson]

This will stay as-is for now. As long as [a b] means concatenation, T[a b] should mean concatenation with an element type.

closing as dup of #7128