MethodErrors when expanding in `Fourier()[:, range]` with non-UnitRange `range`

[DanielVandH]

(jl_RPqcPu) pkg> st
Status `C:\Users\djv23\AppData\Local\Temp\jl_RPqcPu\Project.toml`
  [b30e2e7b] ClassicalOrthogonalPolynomials v0.15.1

julia> using ClassicalOrthogonalPolynomials

julia> expand(Fourier(), sin) # fine
Fourier{Float64}() * [-5.819717467793159e-18, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0  …  ]

julia> expand(Fourier()[:, 1:3], sin) # fine
Fourier{Float64}() * [7.401486830834377e-17, 0.9999999999999998, -7.401486830834377e-17, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0  …  ]

julia> expand(Fourier()[:, 1:2:3], sin) # not fine
ERROR: MethodError: no method matching zero(::Type{Matrix{Int64}})
The function `zero` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  zero(::Type{Union{}}, Any...)
   @ Base number.jl:310
  zero(::Type{Infinities.Infinity})
   @ Infinities C:\Users\djv23\.julia\packages\Infinities\r9wt7\src\Infinities.jl:46
  zero(::Type{Dates.Date})
   @ Dates C:\Users\djv23\.julia\juliaup\julia-1.11.3+0.x64.w64.mingw32\share\julia\stdlib\v1.11\Dates\src\types.jl:459
  ...

Stacktrace:
  [1] getindex_value
    @ C:\Users\djv23\.julia\packages\FillArrays\lVl4c\src\FillArrays.jl:321 [inlined]
  [2] broadcasted(::Base.Broadcast.DefaultArrayStyle{2}, op::typeof(identity), r::FillArrays.Zeros{Matrix{…}, 2, Tuple{…}})
    @ FillArrays C:\Users\djv23\.julia\packages\FillArrays\lVl4c\src\fillbroadcast.jl:79
  [3] broadcasted
    @ .\broadcast.jl:1325 [inlined]
  [4] cache_filldata!(::LazyArrays.CachedArray{…}, ::UnitRange{…}, ::Base.OneTo{…})
    @ LazyArrays C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\cache.jl:230
  [5] resizedata!(::ArrayLayouts.DenseColumnMajor, ::ArrayLayouts.ZerosLayout, ::LazyArrays.CachedArray{…}, ::Int64, ::Int64)
    @ LazyArrays C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\cache.jl:270
  [6] resizedata!(::LazyArrays.CachedArray{Matrix{…}, 2, Matrix{…}, FillArrays.Zeros{…}}, ::Int64, ::Int64)
    @ LazyArrays C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\cache.jl:221
  [7] setindex!
    @ C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\cache.jl:85 [inlined]
  [8] _setindex!
    @ .\abstractarray.jl:1448 [inlined]
  [9] setindex!
    @ .\abstractarray.jl:1413 [inlined]
 [10] collect_to_with_first!(dest::LazyArrays.CachedArray{…}, v1::Matrix{…}, itr::Base.Generator{…}, st::Tuple{…})
    @ Base .\array.jl:826
 [11] collect(itr::Base.Generator{Base.Iterators.ProductIterator{Tuple{…}}, BlockArrays.var"#26#29"{Matrix{…}, Tuple{…}}})
    @ Base .\array.jl:801
 [12] map(f::Function, A::Base.Iterators.ProductIterator{Tuple{BlockArrays.BlockRange{…}, BlockArrays.BlockRange{…}}})
    @ Base .\abstractarray.jl:3399
 [13] initialized_blocks_BlockArray(::Type{Matrix{…}}, baxes::Tuple{BlockArrays.BlockedOneTo{…}, Base.OneTo{…}})
    @ BlockArrays C:\Users\djv23\.julia\packages\BlockArrays\8VbzB\src\blockarray.jl:190
 [14] BlockArray
    @ C:\Users\djv23\.julia\packages\BlockArrays\8VbzB\src\blockarray.jl:201 [inlined]
 [15] similar
    @ C:\Users\djv23\.julia\packages\BlockArrays\8VbzB\src\blockarray.jl:440 [inlined]
 [16] similar(a::LinearAlgebra.Diagonal{Int64, FillArrays.Ones{…}}, dims::Tuple{BlockArrays.BlockedOneTo{…}, Base.OneTo{…}})
    @ Base .\abstractarray.jl:822
 [17] _unsafe_getindex(::IndexCartesian, ::LinearAlgebra.Diagonal{…}, ::Base.Slice{…}, ::StepRange{…})
    @ Base .\multidimensional.jl:927
 [18] _getindex
    @ .\multidimensional.jl:915 [inlined]
 [19] getindex
    @ .\abstractarray.jl:1312 [inlined]
 [20] arguments(::LazyArrays.ApplyLayout{typeof(*)}, V::QuasiArrays.SubQuasiArray{Float64, 2, Fourier{…}, Tuple{…}, false})
    @ ContinuumArrays C:\Users\djv23\.julia\packages\ContinuumArrays\0WcPp\src\bases\bases.jl:639
 [21] arguments(::ContinuumArrays.SubBasisLayout, A::QuasiArrays.SubQuasiArray{Float64, 2, Fourier{…}, Tuple{…}, false})
    @ ContinuumArrays C:\Users\djv23\.julia\packages\ContinuumArrays\0WcPp\src\bases\bases.jl:617
 [22] arguments(a::QuasiArrays.SubQuasiArray{Float64, 2, Fourier{Float64}, Tuple{Inclusion{…}, StepRange{…}}, false})
    @ QuasiArrays C:\Users\djv23\.julia\packages\QuasiArrays\hFvLm\src\lazyquasiarrays.jl:226
 [23] _mul_arguments
    @ C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\linalg\mul.jl:136 [inlined]
 [24] _mul_arguments(::ContinuumArrays.SubBasisLayout, A::QuasiArrays.SubQuasiArray{Float64, 2, Fourier{…}, Tuple{…}, false})
    @ ContinuumArrays C:\Users\djv23\.julia\packages\ContinuumArrays\0WcPp\src\bases\bases.jl:616
 [25] _mul_arguments
    @ C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\linalg\mul.jl:139 [inlined]
 [26] _flatten
    @ C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\linalg\mul.jl:147 [inlined]
 [27] simplify
    @ C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\linalg\mul.jl:345 [inlined]
 [28] simplify(M::ArrayLayouts.Mul{…})
    @ LazyArrays C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\linalg\mul.jl:355
 [29] copy
    @ C:\Users\djv23\.julia\packages\LazyArrays\ltmzk\src\linalg\mul.jl:360 [inlined]
 [30] materialize
    @ C:\Users\djv23\.julia\packages\ArrayLayouts\B2wRU\src\mul.jl:137 [inlined]
 [31] mul
    @ C:\Users\djv23\.julia\packages\ArrayLayouts\B2wRU\src\mul.jl:138 [inlined]
 [32] *(A::QuasiArrays.SubQuasiArray{Float64, 2, Fourier{…}, Tuple{…}, false}, B::Vector{Float64})
    @ QuasiArrays C:\Users\djv23\.julia\packages\QuasiArrays\hFvLm\src\matmul.jl:25
 [33] expand(A::QuasiArrays.SubQuasiArray{Float64, 2, Fourier{…}, Tuple{…}, false}, f::Function)
    @ ContinuumArrays C:\Users\djv23\.julia\packages\ContinuumArrays\0WcPp\src\bases\bases.jl:348
 [34] top-level scope
    @ REPL[17]:1
Some type information was truncated. Use `show(err)` to see complete types.

[DanielVandH]

This seems to be fixed if using the method proposed in JuliaApproximation/HarmonicOrthogonalPolynomials.jl#87 (comment) (proposed to fix piracy, not an issue like this). Maybe that method should be added somewhere to fix both this and JuliaApproximation/HarmonicOrthogonalPolynomials.jl#87. (Doesn't fix #236 though, not sure what the deal is there.)