Out of place `linsolve=KrylovJL_GMRES` is broken

[oscardssmith]

julia> using LinearSolve, OrdinaryDiffEq
julia> function rober(u, p, t)
           y₁, y₂, y₃ = u
           k₁, k₂, k₃ = p
           [-k₁ * y₁ + k₃ * y₂ * y₃,
             k₁ * y₁ - k₃ * y₂ * y₃ - k₂ * y₂^2,
             k₂ * y₂^2]
       end
rober (generic function with 1 method)

julia> prob = ODEProblem(rober, [1.0,0.0,0.0],(0.0,1e3),(0.04,3e7,1e4));

julia> solve(prob, FBDF(linsolve=KrylovJL_GMRES()))
ERROR: MethodError: Cannot `convert` an object of type LinearAlgebra.LU{Float64, Matrix{Float64}, Vector{Int64}} to an object of type OrdinaryDiffEq.WOperator{false, Float64, LinearAlgebra.UniformScaling{Bool}, Float64, Matrix{Float64}, Vector{Float64}, Matrix{Float64}, FunctionOperator{true, true, false, Float64, SparseDiffTools.FwdModeAutoDiffVecProd{SparseDiffTools.JacFunctionWrapper{false, true, 1, OrdinaryDiffEq.var"#790#793", Vector{Float64}, Tuple{Float64, Float64, Float64}, Float64}, Vector{Float64}, Tuple{Vector{ForwardDiff.Dual{ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}, Float64, 1}}, Vector{ForwardDiff.Dual{ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}, Float64, 1}}}, typeof(SparseDiffTools.auto_jacvec), typeof(SparseDiffTools.auto_jacvec!)}, Nothing, Nothing, Nothing, @NamedTuple{islinear::Bool, isconvertible::Bool, isconstant::Bool, opnorm::Nothing, issymmetric::Bool, ishermitian::Bool, isposdef::Bool, isinplace::Bool, outofplace::Bool, has_mul5::Bool, ifcache::Bool, T::DataType, batch::Bool, size::Tuple{Int64, Int64}, sizes::Tuple{Tuple{Int64}, Tuple{Int64}}, accepted_kwargs::Tuple{}, kwargs::Dict{Symbol, Any}}, Tuple{Float64, Float64, Float64}, Float64, Tuple{Vector{Float64}, Vector{Float64}}, Float64, Float64}}

Closest candidates are:
  convert(::Type{T}, ::T) where T
   @ Base Base.jl:84

Stacktrace:
  [1] setproperty!(x::OrdinaryDiffEq.NLNewtonConstantCache{…}, f::Symbol, v::LinearAlgebra.LU{…})
    @ Base ./Base.jl:40
  [2] update_W!(nlsolver::OrdinaryDiffEq.NLSolver{…}, integrator::OrdinaryDiffEq.ODEIntegrator{…}, cache::OrdinaryDiffEq.FBDFConstantCache{…}, dtgamma::Float64, repeat_step::Bool, newJW::Nothing)
    @ OrdinaryDiffEq ~/.julia/dev/OrdinaryDiffEq/src/derivative_utils.jl:837
  [3] update_W!(nlsolver::OrdinaryDiffEq.NLSolver{…}, integrator::OrdinaryDiffEq.ODEIntegrator{…}, cache::OrdinaryDiffEq.FBDFConstantCache{…}, dtgamma::Float64, repeat_step::Bool)
    @ OrdinaryDiffEq ~/.julia/dev/OrdinaryDiffEq/src/derivative_utils.jl:827
  [4] nlsolve!(nlsolver::OrdinaryDiffEq.NLSolver{…}, integrator::OrdinaryDiffEq.ODEIntegrator{…}, cache::OrdinaryDiffEq.FBDFConstantCache{…}, repeat_step::Bool)
    @ OrdinaryDiffEq ~/.julia/dev/OrdinaryDiffEq/src/nlsolve/nlsolve.jl:27
  [5] perform_step!(integrator::OrdinaryDiffEq.ODEIntegrator{…}, cache::OrdinaryDiffEq.FBDFConstantCache{…}, repeat_step::Bool)
    @ OrdinaryDiffEq ~/.julia/dev/OrdinaryDiffEq/src/perform_step/bdf_perform_step.jl:1136
  [6] perform_step!
    @ ~/.julia/dev/OrdinaryDiffEq/src/perform_step/bdf_perform_step.jl:1076 [inlined]
  [7] solve!(integrator::OrdinaryDiffEq.ODEIntegrator{…})
    @ OrdinaryDiffEq ~/.julia/dev/OrdinaryDiffEq/src/solve.jl:551
  [8] __solve(::ODEProblem{…}, ::FBDF{…}; kwargs::@Kwargs{})
    @ OrdinaryDiffEq ~/.julia/dev/OrdinaryDiffEq/src/solve.jl:7
  [9] __solve
    @ ~/.julia/dev/OrdinaryDiffEq/src/solve.jl:1 [inlined]
 [10] #solve_call#44
    @ ~/.julia/dev/DiffEqBase/src/solve.jl:612 [inlined]
 [11] solve_call(_prob::ODEProblem{…}, args::FBDF{…})
    @ DiffEqBase ~/.julia/dev/DiffEqBase/src/solve.jl:569
 [12] solve_up(prob::ODEProblem{…}, sensealg::Nothing, u0::Vector{…}, p::Tuple{…}, args::FBDF{…}; kwargs::@Kwargs{})
    @ DiffEqBase ~/.julia/dev/DiffEqBase/src/solve.jl:1080
 [13] solve_up
    @ ~/.julia/dev/DiffEqBase/src/solve.jl:1066 [inlined]
 [14] solve(prob::ODEProblem{…}, args::FBDF{…}; sensealg::Nothing, u0::Nothing, p::Nothing, wrap::Val{…}, kwargs::@Kwargs{})
    @ DiffEqBase ~/.julia/dev/DiffEqBase/src/solve.jl:1003
 [15] solve(prob::ODEProblem{…}, args::FBDF{…})
    @ DiffEqBase ~/.julia/dev/DiffEqBase/src/solve.jl:993
 [16] top-level scope
    @ REPL[23]:1
Some type information was truncated. Use `show(err)` to see complete types.

This problem exists for all KrylongJL methods, but succeeds for in place problems.

[prbzrg]

Because of its usage in default algorithm, I'm getting it too https://github.com/SciML/OrdinaryDiffEq.jl/blob/6e9cf4055f8fc53f88af8a97df034d50927dd869/src/composite_algs.jl#L114

[ChrisRackauckas]

Yeah now a lot of people might run into this. Try to fix it early next week, or we can remove it from the default alg when out of place if we really have to.