Support operations between NullableArray and Nullable/scalar

[nalimilan]

Basic operations such as this currently don't work:

julia> NullableArray(1:2) + Nullable(1)
ERROR: MethodError: no method matching +(::NullableArrays.NullableArray{Int64,1}, ::Nullable{Int64})
Closest candidates are:
  +(::Any, ::Any, ::Any, ::Any...) at operators.jl:133
  +{S}(::Nullable{Union{}}, ::Nullable{S}) at /home/milan/.julia/NullableArrays/src/operators.jl:130
  +{S,T}(::AbstractArray{S,N}, ::AbstractArray{T,N}) at arraymath.jl:53
  ...

julia> NullableArray(1:2) .+ Nullable(1)
ERROR: MethodError: no method matching .+(::NullableArrays.NullableArray{Int64,1}, ::Nullable{Int64})
Closest candidates are:
  .+{T}(::AbstractArray{T,N}, ::Number) at arraymath.jl:78
  .+(::AbstractArray{T,N}...) at broadcast.jl:304

[johnmyleswhite]

For .+, perhaps worth waiting for 0.6 rather than implementing these in a way that would get replaced by broadcasting in the future? Non-dotted + seems like a special case we have to implement no matter what happens in Base.

[davidagold]

What does waiting for 0.6 have to do with it?

[johnmyleswhite]

Because 0.6 will offer a version of .+ that's totally derived from broadcast. I am not sure, but I think 0.5 still treats the infix dot functions in a special way.

[davidagold]

Ah. Yes it does.

[mkborregaard]

I don't know whether this is the right place to post this, but I am getting some strange behaviour:

julia> 3 < Nullable(4)
true
julia> 3 < Nullable()
false

the scalar boolean return value may be necessary for type stability, but I am thinking this may cause issues (I would expect the last expression to return Null).

[nalimilan]

@mkborregaard That's really weird, as those two calls do not work at all for me. Maybe you've loaded a particular package which overloads these methods?

[mkborregaard]

Hah! yes, Query.jl is the culprit. Should I open an issue over there? EDIT: I'll do that now (https://github.com/davidanthoff/Query.jl/issues/79).

julia> 3 < Nullable(4)
ERROR: MethodError: no method matching isless(::Int64, ::Nullable{Int64})
Closest candidates are:
  isless(::Real, ::AbstractFloat) at operators.jl:41
  isless(::Real, ::Real) at operators.jl:75
  isless(::Integer, ::Char) at deprecated.jl:49
 in <(::Int64, ::Nullable{Int64}) at ./operators.jl:63

julia> using DataFrames

julia> 3 < Nullable(4)
ERROR: MethodError: no method matching isless(::Int64, ::Nullable{Int64})
Closest candidates are:
  isless(::Nullable{Union{}}, ::Nullable{T}) at /Users/michael/.julia/v0.5/NullableArrays/src/operators.jl:161
  isless(::Real, ::AbstractFloat) at operators.jl:41
  isless(::Real, ::Real) at operators.jl:75
  ...
 in <(::Int64, ::Nullable{Int64}) at ./operators.jl:63

julia> using Query
WARNING: Method definition require(Symbol) in module Base at loading.jl:345 overwritten in module Query at /Users/michael/.julia/v0.5/Requires/src/require.jl:12.
WARNING: Method definition ==(Base.Nullable{#T<:Any}, Base.Nullable{Union{}}) in module NullableArrays at /Users/michael/.julia/v0.5/NullableArrays/src/operators.jl:140 overwritten in module Query at /Users/michael/.julia/v0.5/Query/src/operators.jl:8.
WARNING: Method definition ==(Base.Nullable{Union{}}, Base.Nullable{#T<:Any}) in module NullableArrays at /Users/michael/.julia/v0.5/NullableArrays/src/operators.jl:139 overwritten in module Query at /Users/michael/.julia/v0.5/Query/src/operators.jl:9.
WARNING: Method definition ==(Base.Nullable{#T1<:Any}, Base.Nullable{#T2<:Any}) in module NullableArrays at /Users/michael/.julia/v0.5/NullableArrays/src/operators.jl:130 overwritten in module Query at /Users/michael/.julia/v0.5/Query/src/operators.jl:77.

julia> 3 < Nullable(4)
true

[nalimilan]

Yes, we've discussed this previously with @davidanthoff. With Julia 0.6 supporting .< for lifting comparison, I think we should remove all these operators in favor of their element-wise/broadcasting/lifting counterparts.

[mkborregaard]

What type would they return? (just out of curiosity)

[nalimilan]

It already works, and it always returns Nullable{Bool}. That's the only way to be able to return Nullable() when one of the operands is null.

[davidanthoff]

Just to note that this won't work in Query. I need proper predicates in Query that return Bool, not Nullable{Bool}, so Query will just opt out of the whole Nullable story if that is the direction base is taking (unless the folks that are driving these changes to Nullable can convince the base crew to allow Nullable{Bool} in filter expressions in generators, if, while, ?: clauses etc.)