Conversion from SparseMatrixCSC to Matrix uses the same eltype.

[blegat]

When converting a SparseMatrixCSC to a Matrix, the current implementation assumes that it can store the result in a matrix of the same eltype

https://github.com/JuliaLang/julia/blob/989de7903e492aa25f5a888c19b4b8e8a3f9f1f0/stdlib/SparseArrays/src/sparsematrix.jl#L404

However, this does not work of zero of the eltype returns an element of a different type.

MCWE:

struct A end
Base.zero(::Type{A}) = 0
using SparseArrays
S = sparse([1, 2], [1, 2], [A(), A()])
Matrix(S)

output:

ERROR: LoadError: MethodError: Cannot `convert` an object of type Int64 to an object of type A
Closest candidates are:
  convert(::Type{T}, ::T) where T at essentials.jl:123
Stacktrace:
 [1] fill!(::Array{A,2}, ::Int64) at ./array.jl:220
 [2] zeros at ./array.jl:403 [inlined]
 [3] zeros at ./array.jl:400 [inlined]
 [4] Array{T,2} where T(::SparseMatrixCSC{A,Int64}) at /home/blegat/git/julia/usr/share/julia/stdlib/v0.7/SparseArrays/src/sparsematrix.jl:404
...

The error is due to the call zero(A, 2, 2).

This example is a bit artificial but the problem also occurs in JuMP with sparse matrices of VariableRef for which the zero type is AffExpr.

[martinholters]

So the element type should be promote_type(Tv, typeof(zero(Tv)))?
From a brief look, the current code tries to avoid calling zero if the matrix is, in fact, dense. If we want to preserve that, we would need to preserve the element type Tv in the dense case, i.e.

A = length(S) == nnz(S) ? Matrix{Tv}(undef, S.m, S.n) : zeros(promote_type(Tv, typeof(zero(Tv))), S.m, S.n)

This would lead to a type-instability in the rare case that Tv != typeof(zero(Tv)). Not sure whether that's a problem.

[blegat]

So the element type should be promote_type(Tv, typeof(zero(Tv))) ?

Yes, that would work.

This would lead to a type-instability in the rare case that Tv != typeof(zero(Tv)). Not sure whether that's a problem.

In which case does one need to convert a full SparseMatrixCSC that has an eltype that does not implement zero into a Matrix ? Maybe we should just use zeros in all case.

[mbauman]

There's a bigger problem with this premise: eltype(S) is wrong. It should be Union{Tv, typeof(zero(Tv))}. Worse: I don't think we can express that in the subtype relationship to AbstractArray.

[SobhanMP]

from the doc of zero

Get the additive identity element for the type of x

this is just wrong.