Christopher Rackauckas

> The autodiff keyword seems superfluous given the autojacvec = true option: true: the Jacobian is constructed via ForwardDiff.jl I'm not sure that's correct. I think you need autojacvec false...

Indeed this looks like the root issue. Haven't gotten the time to dive in yet though.

Well the way to solve it would be to find the dispatches in the adjoint chain which are looking for `AbstractODEAlgorithm` but not `Nothing`, since the default algorithm is the...

> @ DiffEqCallbacks C:\Users\prbzr\.julia\packages\DiffEqCallbacks\9fKPq\src\integrating_sum.jl:37 That should be using current order instead of order.

Yes that's expected.

Completed, I'll update the test here.

MWE? And no, that still doesn't motivate this PR at all. The nothing dispatch is a default setter. Deleting the default setter is not the solution...

Here is an SDDE method which just came out. https://www.sciencedirect.com/science/article/pii/S0377042718307738 I think it could be implemented in this nested form?

Cool! Let us know if you need any help. StochasticDiffEq.jl has slightly different stepping behavior due to how adaptivity in SDEs works, but it should be doable in about the...

I tried to track this down and I think this is an instance of https://github.com/JuliaLang/julia/issues/35800 . The first case that I saw interpolations having weird allocations that couldn't be fixed...