queryverse / queryverse/DataValues.jl

Should comparison operators return Bool values?

Open
#7 5 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Julia
Stars
14
Forks
6
Avg merge
1d 7h
Merged PRs (30d)
2

Description

From the discussions in https://github.com/JuliaStats/NullableArrays.jl/pull/85, https://github.com/JuliaData/Roadmap.jl/issues/3, and https://github.com/JuliaStats/DataFramesMeta.jl/issues/58 it seems that the main open question is the behavior of comparison operators. I understand that there are good reasons for the current behavior of NAables. Unfortunately, it could lead to subtle errors, e.g.:

julia> include("NAable.jl"); using .NAables

julia> a = [NAable(1), NAable(-1), NAable{Int}()]
3-element Array{NAables.NAable{Int64},1}:
1  
-1 
#NA

julia> positive = broadcast(>=, a, 0)
3-element Array{Bool,1}:
true
false
false

julia> b = Any[NA, NA, NA]; b[positive] = "+"; b[~positive] = "-"; b
3-element Array{Any,1}:
"+"
"-"
"-"

julia> getsign(x) = x < 0 ? "-" : "+"
getsign (generic function with 1 method)

julia> getsign.(a)
3-element Array{String,1}:
"+"
"-"
"+"

If comparison operators return NAable{Bool} instead, and a function like
bool{T<:Bool}(x::NAable{T}) = isna(x) ? false : x.value exists, the above would become:

julia> include("NAable3VL.jl"); using .NAables

julia> a = [NAable(1), NAable(-1), NAable{Int}()]
3-element Array{NAables.NAable{Int64},1}:
1  
-1 
#NA

julia> positive = broadcast(>=, a, 0)
3-element Array{NAables.NAable{Bool},1}:
true 
false
#NA  

julia> b = Any[NA, NA, NA]; b[bool.(positive)] = "+"; b[bool.(~positive)] = "-"; b
3-element Array{Any,1}:
"+"
"-"
NA 

julia> getsign(x) = x < 0 ? "-" : "+"
getsign (generic function with 1 method)

julia> getsign.(a)
ERROR: TypeError: non-boolean (NAables.NAable{Bool}) used in boolean context

Maybe one can automatically treat NA as false in filter-like contexts like Boolean indexing, generators, @where, ..., so one does not have to wrap such conditions in bool(...). But because it is not safe to assume that the author of an arbitrary function using if ... else .. end or the ternary operator has anticipated the possibility of NAs, an error should be thrown if NA occurs in a comparison in a control flow statement. As Stefan Karpinsky pointed out in https://github.com/JuliaStats/NullableArrays.jl/pull/85#issuecomment-240859040, it could still be possible to allow non-NA NAable{Bool}s in such situations.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Read the current behavior illustrated through NAable.jl and the proposed behavior in NAable3VL.jl, then review the linked NullableArrays.jl, Roadmap.jl, and DataFramesMeta.jl discussions. The issue needs a settled decision on comparison results and their behavior in indexing and control flow before implementation can be considered done.

Written by the indexing model from the issue text.

Assessment

Tech stack
julia
Domain
data
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.