vcat on SparseVector's of different eltype returns a SparseMatrixCSC on v0.6-rc2

[blegat]

I would expect the concatenation of sparse vectors to be a sparse vector but it is not the case when their eltype is different

julia> versioninfo()
Julia Version 0.6.0-rc2.0
Commit 68e911b (2017-05-18 02:31 UTC)

julia> vcat(spzeros(2), spzeros(3)) # Same eltype
5-element SparseVector{Float64,Int64} with 0 stored entries

julia> vcat(spzeros(Int, 2), spzeros(3)) # Different eltype
5×1 SparseMatrixCSC{Float64,Int64} with 0 stored entries

This is not a regression since the bug is present in Julia v0.5 too

julia> versioninfo()
Julia Version 0.5.2
Commit f4c6c9d4bb* (2017-05-06 16:34 UTC)

julia> vcat(spzeros(Int, 2), spzeros(Float64, 3))
5×1 sparse matrix with 0 Float64 nonzero entries

EDIT: The following bug is however a regression:

julia> versioninfo()
Julia Version 0.6.0-rc2.0
Commit 68e911b (2017-05-18 02:31 UTC)

julia> vcat(Float64[], spzeros(1))
1×1 SparseMatrixCSC{Float64,Int64} with 0 stored entries

julia> versioninfo()
Julia Version 0.5.2
Commit f4c6c9d4bb* (2017-05-06 16:34 UTC)

julia> vcat(Float64[], spzeros(1))
Sparse vector of length 1 with 0 Float64 nonzero entries:

[ararslan]

Fixed by #22299 and #22301.

[blegat]

Thank you @Sacha0 !