Makie is unable to plot a DimArray when a dimension is constructed from another DimArray

[alex-s-gardner]

See below MWE

[asinghvi17]

How are you saving to disk?

[alex-s-gardner]

OK... I found the problem... da2 had a dimension built from a DimArray... I wonder if there is an issue here as a DimArray should behave as an AbstractArray. Here's the MWE:

This plots just fine:

julia> date = [DateTime("1985-03-13T00:00:00"), DateTime("1985-04-12T00:00:00"), DateTime("1985-05-12T00:00:00"), DateTime("1985-06-11T00:00:00")]
4-element Vector{DateTime}:
 1985-03-13T00:00:00
 1985-04-12T00:00:00
 1985-05-12T00:00:00
 1985-06-11T00:00:00

julia> ddate = Dim{:date}(date)
date [DateTime("1985-03-13T00:00:00"), …, DateTime("1985-06-11T00:00:00")]

julia> data = rand(length(ddate))
4-element Vector{Float64}:
 0.2693751970756716
 0.6519072128117327
 0.9912748870960395
 0.4505443280597703

julia> da1 = DimArray(data, ddate)
┌ 4-element DimArray{Float64, 1} ┐
├────────────────────────────────┴─────────────────────────────────────────────────────────────────────────────────────────── dims ┐
  ↓ date Sampled{DateTime} [DateTime("1985-03-13T00:00:00"), …, DateTime("1985-06-11T00:00:00")] ForwardOrdered Irregular Points
└──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘
 1985-03-13T00:00:00  0.269375
 1985-04-12T00:00:00  0.651907
 1985-05-12T00:00:00  0.991275

This throws an error when the dimension is constructed from a DimArray:

julia> date = [DateTime("1985-03-13T00:00:00"), DateTime("1985-04-12T00:00:00"), DateTime("1985-05-12T00:00:00"), DateTime("1985-06-11T00:00:00")]
4-element Vector{DateTime}:
 1985-03-13T00:00:00
 1985-04-12T00:00:00
 1985-05-12T00:00:00
 1985-06-11T00:00:00

julia> date2 = DimArray(date, ddate)
┌ 4-element DimArray{DateTime, 1} ┐
├─────────────────────────────────┴────────────────────────────────────────────────────────────────────────────────────────── dims ┐
  ↓ date Sampled{DateTime} [DateTime("1985-03-13T00:00:00"), …, DateTime("1985-06-11T00:00:00")] ForwardOrdered Irregular Points
└──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘
 1985-03-13T00:00:00  1985-03-13T00:00:00
 1985-04-12T00:00:00  1985-04-12T00:00:00
 1985-05-12T00:00:00  1985-05-12T00:00:00
 1985-06-11T00:00:00  1985-06-11T00:00:00

julia> ddate2 = Dim{:date}(date2)
date [DateTime("1985-03-13T00:00:00"), …, DateTime("1985-06-11T00:00:00")]

julia> da2 = DimArray(data, ddate2)
┌ 4-element DimArray{Float64, 1} ┐
├────────────────────────────────┴─────────────────────────────────────────────────────────────────────────────────────────── dims ┐
  ↓ date Sampled{DateTime} [DateTime("1985-03-13T00:00:00"), …, DateTime("1985-06-11T00:00:00")] ForwardOrdered Irregular Points
└──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘
 1985-03-13T00:00:00  0.269375
 1985-04-12T00:00:00  0.651907
 1985-05-12T00:00:00  0.991275
 1985-06-11T00:00:00  0.450544

julia> lines(da2)
ERROR: ArgumentError:     Conversion failed for Lines (With conversion trait PointBased()) with args: Tuple{Vector{Int64}, Vector{Point{2, Float64}}} .
    Lines requires to convert to argument types Tuple{AbstractVector{<:Union{Point2, Point3}}}, which convert_arguments didn't succeed in.
    To fix this overload convert_arguments(P, args...) for Lines or PointBased() and return an object of type Tuple{AbstractVector{<:Union{Point2, Point3}}}.`

Stacktrace:
 [1] conversion_pipeline(P::Type{…}, used_attrs::Tuple{}, args::Tuple{…}, args_obs::Tuple{…}, user_attributes::Dict{…}, deregister::Vector{…}, recursion::Int64)
   @ Makie ~/.julia/packages/Makie/8h0bl/src/interfaces.jl:241
 [2] conversion_pipeline(P::Type{…}, used_attrs::Tuple{}, args::Tuple{…}, args_obs::Tuple{…}, user_attributes::Dict{…}, deregister::Vector{…}, recursion::Int64)
   @ Makie ~/.julia/packages/Makie/8h0bl/src/interfaces.jl:234
 [3] conversion_pipeline(P::Type{…}, used_attrs::Tuple{}, args::Tuple{…}, args_obs::Tuple{…}, user_attributes::Dict{…}, deregister::Vector{…})
   @ Makie ~/.julia/packages/Makie/8h0bl/src/interfaces.jl:214
 [4] (Lines)(user_args::Tuple{DimVector{…}, Vector{…}}, user_attributes::Dict{Symbol, Any})
   @ Makie ~/.julia/packages/Makie/8h0bl/src/interfaces.jl:272
 [5] _create_plot(::Function, ::Dict{…}, ::DimVector{…}, ::Vararg{…})
   @ Makie ~/.julia/packages/Makie/8h0bl/src/figureplotting.jl:316
 [6] #lines#60
   @ ~/.julia/packages/MakieCore/hXATT/src/recipes.jl:436 [inlined]
 [7] lines(A::DimVector{…}; axislegendkw::@NamedTuple{}, axislegend::@NamedTuple{}, axis::@NamedTuple{}, attributes::@Kwargs{})
   @ DimensionalDataMakie ~/.julia/packages/DimensionalData/5jhQ2/ext/DimensionalDataMakie.jl:78
 [8] lines(A::DimVector{…})
   @ DimensionalDataMakie ~/.julia/packages/DimensionalData/5jhQ2/ext/DimensionalDataMakie.jl:74
 [9] top-level scope
   @ REPL[79]:1
Some type information was truncated. Use `show(err)` to see complete types.

[alex-s-gardner]

Note that the display() of da1 and da2 are identical

[asinghvi17]

Aha interesting! Yes it probably hits a second expansion somehow.

But presumably dims should be treated as strictly normal base arrays. Maybe we need to check if they are dimarrays and extract .data?

[rafaqz]

Yeah probably we should do a convert(Array, parent(lookup)) rather than just parent(lookup) so that there are no further expansions.