some issues with sparse matrix operations combinations: slice, sum and assignment

[alsam]

While trying some sparse matrix operations stumbled upon

julia> y = spzeros(7,7)
7x7 sparse matrix with 0 Float64 entries:

julia> y[2,1] = y[2,3] = y[2,5] = 3
3
julia> z = spzeros(7,7)
7x7 sparse matrix with 0 Float64 entries:

julia> z = -y
7x7 sparse matrix with 3 Float64 entries:
    [2, 1]  =  -3.0
    [2, 3]  =  -3.0
    [2, 5]  =  -3.0

Ok, so far, so good

julia> for i = 1:7
       z[i,i] = sum(y[i,:])
       end
ERROR: no method convert(Type{Float64}, Array{Float64,1})
 in setindex! at sparse/sparsematrix.jl:969
 in anonymous at no file:2

but these operations applied separately are good:

julia> for i = 1:7
       z[i,i] = 7.0
       end

julia> for i = 1:7
       s = sum(y[i,:])
       end

Delving into it:

julia> typeof(sum(y[2,:]))

Array{Float64,1}

julia> sum(y[2,:])

1-element Array{Float64,1}:
 9.0

julia> sum([1,2,3])
6

julia> sum([1,2,3][1:2])
3

julia> typeof(sum([1,2,3][1:2]))
Int64

So why for sparse vector sum produced 1-element Array instead of a scalar value?

Thanks!

[jiahao]

I could not reproduce the problem. Which version of Julia are you using?

[alsam]

I confirm that now it really works as expected with up-to-date Julia.

julia> y = spzeros(7,7)
7x7 sparse matrix with 0 Float64 entries:

julia> y[2,1] = y[2,3] = y[2,5] = 3
3

julia> z = -y
7x7 sparse matrix with 3 Float64 entries:
        [2, 1]  =  -3.0
        [2, 3]  =  -3.0
        [2, 5]  =  -3.0

julia> for i = 1:7
       z[i,i] = sum(y[i,:])
       end

julia> z
7x7 sparse matrix with 4 Float64 entries:
        [2, 1]  =  -3.0
        [2, 2]  =  9.0
        [2, 3]  =  -3.0
        [2, 5]  =  -3.0

So the issue was resolved.
Thanks!