3.4.1 -> 3.4.2 breaking

[j-fu]

Some methods for AbstracDiffEqArray are broken since 3.4.2
Before, it was possible to calculate e.g. a maximum of an instance
of a subtype of AbstractDiffEqArray.

** MRE: **

using RecursiveArrayTools
using Test

abstract type AbstractTransientSolution{T, N, A, B} <: AbstractDiffEqArray{T, N, A} end

mutable struct TransientSolution{T, N, A, B} <: AbstractTransientSolution{T, N, A, B}
    u::A
    t::B
end

function TransientSolution(vec::AbstractVector{T}, ts, ::NTuple{N}) where {T, N}
    TransientSolution{T, N, typeof(vec), typeof(ts)}(vec, ts)
end

TransientSolution(vec::AbstractVector, ts::AbstractVector) = TransientSolution(vec, ts, (size(vec[1])..., length(vec)))

times=range(0,1,length=11)
tsol1d=TransientSolution([rand(20) for t in times],times)
tsol2d=TransientSolution([rand(2,20) for t in times],times)

@testset "mwe" begin
    @test maximum(tsol1d)>0
    @test maximum(tsol2d)>0
end

** Error for first test: **

Test threw exception
Expression: maximum(tsol1d) > 0
ArgumentError: number of indices (1) must match the parent dimensionality (2)
Stacktrace:
[1] check_parent_index_match(parent::TransientSolution{Vector{Float64}, 2, Vector{Vector{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}, ::Tuple{Bool})
@ Base ./subarray.jl:43
[2] check_parent_index_match(parent::TransientSolution{Vector{Float64}, 2, Vector{Vector{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}, indices::Tuple{Base.Slice{Base.OneTo{Int64}}})
@ Base ./subarray.jl:41
[3] SubArray
@ ./subarray.jl:21 [inlined]
[4] SubArray
@ ./subarray.jl:32 [inlined]
[5] view
@ ~/.julia/packages/RecursiveArrayTools/SkCK5/src/vector_of_array.jl:514 [inlined]
[6] mapreduce(f::Function, op::Function, A::TransientSolution{Vector{Float64}, 2, Vector{Vector{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}; kwargs::@Kwargs{})
@ RecursiveArrayTools ~/.julia/packages/RecursiveArrayTools/SkCK5/src/vector_of_array.jl:689
[7] mapreduce(f::Function, op::Function, A::TransientSolution{Vector{Float64}, 2, Vector{Vector{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}})
@ RecursiveArrayTools ~/.julia/packages/RecursiveArrayTools/SkCK5/src/vector_of_array.jl:688
[8] maximum(a::TransientSolution{Vector{Float64}, 2, Vector{Vector{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}; kw::@Kwargs{})
@ Base ./reduce.jl:764

** Error for second test: **

Test threw exception
Expression: maximum(tsol2d) > 0
UndefVarError: `J` not defined
Stacktrace:
[1] view
@ ~/.julia/packages/RecursiveArrayTools/SkCK5/src/vector_of_array.jl:513 [inlined]
[2] mapreduce(f::Function, op::Function, A::TransientSolution{Matrix{Float64}, 3, Vector{Matrix{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}; kwargs::@Kwargs{})
@ RecursiveArrayTools ~/.julia/packages/RecursiveArrayTools/SkCK5/src/vector_of_array.jl:689
[3] mapreduce(f::Function, op::Function, A::TransientSolution{Matrix{Float64}, 3, Vector{Matrix{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}})
@ RecursiveArrayTools ~/.julia/packages/RecursiveArrayTools/SkCK5/src/vector_of_array.jl:688
[4] maximum(a::TransientSolution{Matrix{Float64}, 3, Vector{Matrix{Float64}}, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}; kw::@Kwargs{})
@ Base ./reduce.jl:764

Environment (please complete the following information):

• Output of using Pkg; Pkg.status()

⌃ [731186ca] RecursiveArrayTools v3.4.2
   [8dfed614] Test

• Output of using Pkg; Pkg.status(; mode = PKGMODE_MANIFEST)

Status `~/Wias/work/julia/dev/VoronoiFVM/eee/Manifest.toml`
[79e6a3ab] Adapt v4.0.1
[4fba245c] ArrayInterface v7.7.0
[9a962f9c] DataAPI v1.16.0
[e2d170a0] DataValueInterfaces v1.0.0
[ffbed154] DocStringExtensions v0.9.3
[46192b85] GPUArraysCore v0.1.6
[82899510] IteratorInterfaceExtensions v1.0.0
[bac558e1] OrderedCollections v1.6.3
[aea7be01] PrecompileTools v1.2.0
[21216c6a] Preferences v1.4.1
[3cdcf5f2] RecipesBase v1.3.4
⌃ [731186ca] RecursiveArrayTools v3.4.2
[ae029012] Requires v1.3.0
[1e83bf80] StaticArraysCore v1.4.2
[2efcf032] SymbolicIndexingInterface v0.3.3
[3783bdb8] TableTraits v1.0.1
[bd369af6] Tables v1.11.1
[56f22d72] Artifacts
[2a0f44e3] Base64
[ade2ca70] Dates
[b77e0a4c] InteractiveUtils
[76f85450] LibGit2
[8f399da3] Libdl
[37e2e46d] LinearAlgebra
[56ddb016] Logging
[d6f4376e] Markdown
[ca575930] NetworkOptions v1.2.0
[de0858da] Printf
[9a3f8284] Random
[ea8e919c] SHA v0.7.0
[9e88b42a] Serialization
[2f01184e] SparseArrays v1.10.0
[10745b16] Statistics v1.10.0
[4607b0f0] SuiteSparse
[fa267f1f] TOML v1.0.3
[8dfed614] Test
[cf7118a7] UUIDs
[4ec0a83e] Unicode
[e66e0078] CompilerSupportLibraries_jll v1.0.5+1
[e37daf67] LibGit2_jll v1.6.4+0
[29816b5a] LibSSH2_jll v1.11.0+1
[c8ffd9c3] MbedTLS_jll v2.28.2+1
[4536629a] OpenBLAS_jll v0.3.23+2
[bea87d4a] SuiteSparse_jll v7.2.1+1
[8e850b90] libblastrampoline_jll v5.8.0+1

• Output of versioninfo()

Julia Version 1.10.0
Commit 3120989f39b (2023-12-25 18:01 UTC)
Build Info:
Official https://julialang.org/ release
Platform Info:
OS: Linux (x86_64-linux-gnu)
CPU: 20 × 13th Gen Intel(R) Core(TM) i7-13700H
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-15.0.7 (ORCJIT, goldmont)
Threads: 1 on 20 virtual cores
Environment:
LD_LIBRARY_PATH = /home/fuhrmann/local/lib64:/home/fuhrmann/local/lib:/usr/local/cuda-12.3/lib64:.:/usr/local/lib64:/usr/local/lib:/home/fuhrmann/local/lib:/home/fuhrmann/Sys/share/pardiso
JULIA_PKG_DEVDIR = /home/fuhrmann/Wias/work/julia/dev
JULIA_PKG_SERVER_REGISTRY_PREFERENCE = eager
JULIA_PROJECT = .
JULIA_VERSION = 1.10.0
JULIA_HISTORY = /home/fuhrmann/.julia_histories/_home_fuhrmann_Wias_work_julia_dev_VoronoiFVM_eee

Additional context

Things are still working on 3.4.1

[ChrisRackauckas]

@AayushSabharwal can you look into these?

[AayushSabharwal]

The types here are wrong. tsol1d should be an AbstractDiffEqArray{Float64, 2, Vector{Vector{Float64}}} but it is a AbstractDiffEqArray{Vector{Float64}, 2, Vector{Vector{Float64}}}. The first type argument is the type of the numbers inside the DiffEqArray, not the type of the subarrays. This constructor fixes the issue:

function TransientSolution(vec::AbstractVector{T}, ts, ::NTuple{N}) where {T, N}
    TransientSolution{eltype(T), N, typeof(vec), typeof(ts)}(vec, ts)
end

[j-fu]

Ok but what about tsol2d ? That fails as well, but for different reason.
IMHO because there is a missing else branch in this if statement:

RecursiveArrayTools.jl/src/vector_of_array.jl

Line 550 in 80f2c86

if length(I) == 1

BTW that is my main case.

[AayushSabharwal]

It's the same problem. If the type is fixed, it won't ever dispatch to that function. Technically there's a missing case, but it only ever occurs when this method is incorrectly dispatched to.

[j-fu]

Ah ok, thank you, this fixes the MRE. Tested in the wild now as well, works.