Restore OrdinaryDiffEqNonlinearSolve solver-author public API#3946
Closed
singhharsh1708 wants to merge 1 commit into
Closed
Restore OrdinaryDiffEqNonlinearSolve solver-author public API#3946singhharsh1708 wants to merge 1 commit into
singhharsh1708 wants to merge 1 commit into
Conversation
SciML#3832 dropped build_nlsolver, nlsolve!, nlsolvefail, compute_step!, initial_η, markfirststage!, du_alias_or_new, anderson, anderson! from the public block, keeping only the three algorithm types. Downstream solver sublibraries (PDIRK, StabilizedIRK, IMEXMultistep, FIRK, DelayDiffEq, StochasticDiffEqImplicit/Leaping) still import these names directly, so their all_explicit_imports_are_public QA check fails on every one of them. Restores exactly the removed set (git blame e1db9b0, "Split public and developer API docs") plus the type.jl docstring cross-reference it also dropped. HomotopyNonlinearSolveAlg, added after SciML#3832, is unaffected and stays in the block. No open PR currently covers this piece: SciML#3940 restores OrdinaryDiffEqDifferentiation's public block (a different owner), and the two prior attempts at the full set (SciML#3872, SciML#3874) were closed without merging.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bug
#3832("Split public and developer API docs") removed most ofOrdinaryDiffEqNonlinearSolve's public block, keeping only the three algorithm types (NLNewton,NLFunctional,NLAnderson, plusHomotopyNonlinearSolveAlgadded later). But the solver-author extension API it dropped —build_nlsolver,nlsolve!,nlsolvefail,compute_step!,initial_η,markfirststage!,du_alias_or_new,anderson,anderson!— is still directly imported by every downstream solver sublibrary that implements an implicit method:OrdinaryDiffEqPDIRK,OrdinaryDiffEqStabilizedIRK,OrdinaryDiffEqIMEXMultistep,OrdinaryDiffEqFIRK,DelayDiffEq,StochasticDiffEqImplicit,StochasticDiffEqLeaping. Theirall_explicit_imports_are_publicQA check fails on every one of them as a result.Fix
Restore exactly the set
#3832removed (confirmed viagit blameone1db9b0), plus thetype.jldocstring cross-reference toanderson/anderson!it also stripped.HomotopyNonlinearSolveAlg, added after#3832in #3876, is untouched.Scope note
This is a different owning package from #3940 (which restores
OrdinaryDiffEqDifferentiation's block) — not a duplicate. No open PR currently covers this piece; the two prior attempts at the full set (#3872, #3874) were closed without merging.Verification
ExplicitImports.check_all_explicit_imports_are_public(OrdinaryDiffEqPDIRK)on Julia 1.12.6 (matching CI): all names from this restoration (build_nlsolver,markfirststage!,nlsolve!,nlsolvefail,du_alias_or_new) no longer flagged. The one remaining flag (_unwrap_val, owned by SciMLBase) is unrelated to this change.qa.jlfor these 9 names in anignore/ei_brokenlist — none has one, so restoring them can't flip anything to an unexpected-pass failure.TRBDF2()solve succeeds (Successretcode) against a full local dev-path checkout.publicdeclaration cannot change runtime dispatch or behavior — it is visibility/tooling-only — so this carries no runtime regression risk by construction.AI Disclosure
Claude assisted with this work.