Christopher Rackauckas

> @ChrisRackauckas I don't follow, what's the error with non-fully initialized types? Julia allows for a type to be partially defined by having an inner type constructor which only puts...

All of the pieces are in now, Enzyme now works on the explicit solvers!

This is already documented: https://github.com/SciML/OrdinaryDiffEq.jl/blob/master/CONTRIBUTING.md#developing-locally. > I do think we want to think pretty carefully about what our plan for dependency management of the sub-packages is. I think we likely...

That's a good idea, I didn't know you could do that.

Yeah that branch is trivial, seems like a mistake. There's a few things with the extrapolation methods that are a bit odd right now. They are one of the best...

> However, the W property does not appear to be utilized in the implementation. It's somewhat implicit. If someone uses `autodiff=false`, the non-W methods can have some deterioration at lower...

Looks almost good to go! Left a few comments.

Rebase

I really want to have Enzyme as default forward mode in v7 as well? But current discussions with @wsmoses and @gdalle suggest it may not make it, as this is...

I would also like to see https://github.com/SciML/OrdinaryDiffEq.jl/issues/1570 completed to the point that we can change defaults to nonlinearsolve, but we need to make sure that doesn't have a performance regression...