Sundials 4.12 and newer break KiteModels

[ufechner7]

My package https://github.com/ufechner7/KiteModels.jl works fine with Sundials 4.11.4 and older, but 4.12.0 and newer make it fail. In addition pre-compilation of Sundials 4.11.4 only works with DiffEqBase = "<6.116.0".

The following unit test shows the problem:

# unittest for Sundials.jl, works fine with 4.11.4 and older
using StaticArrays, Sundials, Test

const KVec3    = MVector{3, Float64}

Base.@kwdef mutable struct KPS4{S, T, P}
    v_apparent::T =       zeros(S, 3)
end

const kps4 = KPS4{Float64, KVec3, 6+4+1}()

function residual!(res, yd, y::MVector{S, Float64}, s::KPS4, time) where S
end

function init_sim!(t_end=1.0)
    y0 = MVector{62, Float64}([13.970413450119487, 0.0, 21.238692070636343, 27.65581376097752, 0.0, 42.66213714321849, 40.976226230518435, 0.0, 64.314401166278, 53.87184032029182, 0.0, 86.22231803750196, 66.28915240374937, 0.0, 108.4048292516046, 78.17713830204762, 0.0, 130.87545423106485, 79.56930502428155, 0.0, 135.70836376062155, 80.90383289255747, 0.0, 137.7696816741141, 80.60126812407692, 2.4016533873456325, 135.3023287520457, 80.60126812407692, -2.4016533873456325, 135.3023287520457, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 150.0, 0.0])
    yd0= MVector{62, Float64}([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0, -9.81, 0.0, 0.0])
    differential_vars = ones(Bool, length(y0))
    solver  = IDA(linear_solver=Symbol("GMRES"), max_order = 4)
    tspan   = (0.0, t_end) 
    abstol  = 0.0006 # max error in m/s and m
    prob    = DAEProblem(residual!, yd0, y0, tspan, kps4, differential_vars=differential_vars)
    integrator = Sundials.init(prob, solver, abstol=abstol, reltol=0.001)
end

@testset "test_simulate        " begin
    STEPS = 500
    integrator = init_sim!()
end
nothing

[ViralBShah]

Is this still an issue?

[ChrisRackauckas]

I'm not sure we ever expected MArray initial conditions to work with Sundials. It's kind of interesting that it did work at one point through auto-conversions, but since Sundials needs an array pointer to work internally on the values there were just implicit conversions and so it wasn't any faster. Now that we've constrained the package to more clearly require Array{Float64}, this "broke", but is really making it front-and-center that it always required Array{Float64}. To get the equivalent behavior as before, one would need to convert in the f function which is fine. I understand it was quite convenient that it worked before, but I think exposing the solver's true usage makes the performance aspect a lot more clear and so I'll say it was an accident this ever worked without being explicit, and it's not a bug that it's now an error.

[ufechner7]

What I don't like is that you say that you closed this as completed, because that is not true. You closed it as "Won't fix". Is this a limitation of Github that you cannot close a bug as "Won't fix"?

[ChrisRackauckas]

Just an oversight.