SciML / NonlinearSolve.jl

High-performance and differentiation-enabled nonlinear solvers (Newton methods), bracketed rootfinding (bisection, Falsi), with sparsity and Newton-Krylov support.
https://docs.sciml.ai/NonlinearSolve/stable/
MIT License
216 stars 39 forks source link

Remove recompile_invalidations #450

Closed ChrisRackauckas closed 1 week ago

ChrisRackauckas commented 1 week ago

@recompile_invalidations should only be used in very specific scenarios, and this is not one of those scenarios. Also, there are big changes being done with https://github.com/SciML/CommonWorldInvalidations.jl. With that, we only need to @recompile_invalidations on a few entry points. In particular, Static.jl, Symbolics.jl, and preferably ForwardDiff.jl and StaticArrays.jl would adopt it too. But this means that in order to handle all of this effectively, in SciML we only need to apply it on Static.jl, Symbolics.jl, and SciMLBase.jl and the whole ecosystem should be fine.

In any case, this library doesn't need it. It shouldn't make a tangible difference in compile times, while it increases precompile times by a lot.