Simplify :alg extraction

[prbzrg]

No description provided.

[prbzrg]

There was a CI failing related to "Steady State Adjoint".
https://github.com/SciML/DiffEqBase.jl/actions/runs/6437743171/job/17483353413#step:6:1071

Based on
https://github.com/SciML/SciMLBase.jl/blob/master/src/deprecated.jl
I replaced types, maybe this fix it.

[prbzrg]

Why doesn't this test have an alg? also it doesn't have DifferentialEquations!
https://github.com/SciML/SciMLSensitivity.jl/blob/282c24ef282d50d431913bfac2a83b025d34ae76/test/steady_state.jl#L411

[ChrisRackauckas]

It's just using the default.

[prbzrg]

It's just using the default.

how does it access to defaults when DifferentialEquations isn't imported?

https://github.com/SciML/SciMLSensitivity.jl/blob/master/test/steady_state.jl

[ChrisRackauckas]

The default is in the kwarg: https://github.com/SciML/SciMLSensitivity.jl/blob/282c24ef282d50d431913bfac2a83b025d34ae76/test/steady_state.jl#L397

[ChrisRackauckas]

@avik-pal do you know why this works in SciMLSensitivity but not here? https://github.com/SciML/DiffEqBase.jl/actions/runs/6455307402/job/17563616620?pr=945#step:6:1405 . Is this something solved in SciML/SciMLSensitivity.jl#905 or did it break here?

[avik-pal]

I fixed it in 2f0a9a4

[avik-pal]

Are nonlinear solvers AbstractDEAlgorithm?

DiffEqBase.jl/src/solve.jl

Line 1460 in dc0e5c1

elseif solve_args[1] isa AbstractDEAlgorithm

[prbzrg]

julia> supertypes(DynamicSS)
(DynamicSS, SteadyStateDiffEq.SteadyStateDiffEqAlgorithm, SciMLBase.AbstractSteadyStateAlgorithm, SciMLBase.AbstractDEAlgorithm, SciMLBase.AbstractSciMLAlgorithm, Any)

[avik-pal]

Yeah but here the dispatch is failing for the discrete solvers.

julia> supertypes(NewtonRaphson)
(NewtonRaphson, NonlinearSolve.AbstractNewtonAlgorithm, NonlinearSolve.AbstractNonlinearSolveAlgorithm, SciMLBase.AbstractNonlinearAlgorithm, SciMLBase.AbstractSciMLAlgorithm, Any)

Dispatch on AbstractSciMLAlgorithm?

[prbzrg]

Is it intended to ignore types of EnsembleAlgorithm and use nothing in

DiffEqBase.jl/src/solve.jl

Lines 1050 to 1051 in 301dd29

	if isempty(args) || length(args) == 1 && typeof(args[1]) <: EnsembleAlgorithm
	__solve(prob, nothing, args...; kwargs...)

?

[ChrisRackauckas]

That looks to me like a bug . It would be good to try and find a code that hits that branch.

[ChrisRackauckas]

this is dropping the rest of the args?

[prbzrg]

No, there isn't any other args, the length check assures us.

[ChrisRackauckas]

oh I see

[ChrisRackauckas]

Can you add tests for the cases which were fixed?

[prbzrg]

Most of the errors were caused by simplifying the alg extractions, and some of them needed changes elsewhere. like adding kwargs to EnsembleProblem or having a default isadaptive for non-AbstractDEAlgorithm. Here, I just imitated previous behaviour but IMO, the needed tests should happen elsewhere. like in SciMLBase.