Error when creating SumType in an untyped vector

[Tortar]

Let's say you have

using SumTypes

@sum_type A{X,Y} begin
    B{X}(x::X)
    C{Y}(y::Y)
end

it seems to work fine

julia> B(1)
B(1)::A{Int64, Uninit}

julia> [B(i) for i in 1:2]
2-element Vector{A{Int64, Uninit}}:
 B(1)::A{Int64, Uninit}
 B(2)::A{Int64, Uninit}

but instead this errors

julia> [B(1)]
ERROR: MethodError: convert(::Type{A{Int64, Uninit}}, ::A{Int64, Uninit}) is ambiguous.

Candidates:
  convert(::Type{var"##_T#226"}, x::var"##_T#226") where {X, Y, var"##_T#226"<:A{X, Y}}
    @ Main ~/.julia/dev/SumTypes/src/sum_type.jl:212
  convert(::Type{<:A{X, Y}}, x::A{X, Uninit}) where {X, Y}
    @ Main ~/.julia/dev/SumTypes/src/sum_type.jl:203
  convert(::Type{A{X, Y}}, x::A{X, Uninit}) where {X, Y}
    @ Main ~/.julia/dev/SumTypes/src/sum_type.jl:202

Possible fix, define
  convert(::Type{A{X, Uninit}}, ::A{X, Uninit}) where X

Stacktrace:
 [1] __inbounds_setindex!(A::Vector{A{Int64, Uninit}}, x::A{Int64, Uninit}, i1::Int64)
   @ Base ./array.jl:1026
 [2] vect(X::A{Int64, Uninit})
   @ Base ./array.jl:165
 [3] top-level scope
   @ REPL[17]:1

Instead if you type the vector it works:

julia> A[B(1)]
1-element Vector{A}:
 B(1)::A{Int64, Uninit}

julia> A{Int}[B(1)]
1-element Vector{A{Int64}}:
 B(1)::A{Int64, Uninit}

[MasonProtter]

Ah crap. There's a lot of convert methods spat out by @sum_type, trying to do various useful things. I'll have to hunt down a way to fix this ambiguity. :(

[Tortar]

Seems related JuliaDynamics/LightSumTypes.jl#15