QA: run_qa v1.6 form + ExplicitImports#1080
Merged
ChrisRackauckas merged 3 commits intoJul 3, 2026
Merged
Conversation
Convert the hand-rolled test/qa/{qa__aqua.jl, qa__explicitimports.jl} to a single
test/qa/qa.jl using SciMLTesting's run_qa v1.6 form with explicit_imports = true.
- Aqua: undefined_exports/persistent_tasks disables preserved; the historical
hard ambiguities failure (PINOODE PDETimeSeriesSolution callable, red on
master pre-conversion) is tracked as aqua_broken = (:ambiguities,) referencing
SciML#1079 (no silent skip; tracked @test_broken placeholder).
- ExplicitImports: all 6 checks enabled. Fixed the unwrap owner-violation by
importing it from its owner SymbolicUtils instead of Symbolics. Remaining
findings are other packages' not-yet-public names (qualified accesses) and the
LuxCore initialparameters/initialstates/parameterlength method-extension
imports, curated into per-check ignore-lists documenting their source packages.
- QA Project.toml: SciMLTesting compat -> "1.6"; drop ExplicitImports (transitive
via SciMLTesting); keep Aqua (the ambiguities sub-check's child process needs
it as a direct dep).
Verified locally vs released SciMLTesting 1.6.0 (clean resolve, no dev-from-branch):
QA group = 14 Pass, 1 Broken (ambiguities, SciML#1079), 0 Fail/Error.
Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
SciMLTesting 1.7 skips the two public-API ExplicitImports checks
(check_all_qualified_accesses_are_public,
check_all_explicit_imports_are_public) on the lts, and the released
base libs (SciMLBase 3.24.0, Symbolics 7.28.1, SymbolicUtils 4.36.0,
ModelingToolkit 11.29.0, ArrayInterface 7.26.0, CommonSolve 0.2.9)
now declare most of the previously-ignored names public. Empty both
public-API ignore-lists, then restore only the names still flagged on
Julia 1.12 (where these checks run): 17 in
all_qualified_accesses_are_public, 0 in all_explicit_imports_are_public
(whole list removed).
Surviving (still non-public on 1.12), by source:
SciMLBase: AbstractDAEAlgorithm, AbstractDAEProblem,
AbstractDiscretizationMetadata, AbstractSDEProblem, __solve,
allowscomplex, calculate_solution_errors!, has_analytic, interp_summary
SymbolicUtils: _iszero; Symbolics: variables
ForwardDiff: derivative, jacobian
QuasiMonteCarlo: generate_design_matrices, sample
Base: mapany; Base.Broadcast: dottable
Verified QA group green-or-broken (0 hard FAIL) on Julia 1.12
(14 Pass, 1 Broken) and Julia 1.10 (12 Pass, 1 Broken; public-API
checks skipped) against registered releases. via_owners / aqua_broken
ignores untouched.
Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
After the SciMLBase 3.27.0 make-public release, four names previously listed in the public-API `all_qualified_accesses_are_public` ignore are now public (verified via `Base.ispublic`) and are dropped: AbstractDAEProblem, AbstractSDEProblem, AbstractDAEAlgorithm, allowscomplex. The remaining 13 names are still non-public against the released versions (SciMLBase: AbstractDiscretizationMetadata/__solve/has_analytic/ interp_summary/calculate_solution_errors!; SymbolicUtils _iszero; Symbolics variables; ForwardDiff derivative/jacobian; QuasiMonteCarlo generate_design_matrices/sample; Base mapany; Base.Broadcast dottable) and stay in the ignore list. No DiffEqBase->SciMLBase source migration is needed: src/ext access no DiffEqBase names. Bump the qa env SciMLTesting compat 1.6 -> 1.7 to admit the release that gates the public-API EI checks to Julia >=1.11. Verified: QA group green on Julia 1.12 against the registered releases (14 pass / 1 broken[aqua ambiguities, pre-existing SciML#1079] / 0 fail-error); public-API checks correctly skip on Julia 1.10 (SciMLTesting >=1.11 gate). Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
ChrisRackauckas-Claude
marked this pull request as ready for review
June 29, 2026 09:51
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
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.
Ignore until reviewed by @ChrisRackauckas.
Brings this repo's QA onto the SciMLTesting
run_qav1.6 form with ExplicitImports enabled (explicit_imports = true). Converts the hand-rolledtest/qa/{qa__aqua.jl, qa__explicitimports.jl}into a singletest/qa/qa.jl.Aqua
undefined_exports = false(ModelingToolkit@reexports an undefinedAbstractDynamicOptProblem) andpersistent_tasks = false(SymbolicsPreallocationToolsExthas__precompile__(false)).ambiguitiessub-check was a hard red onmasterbefore this PR (QA (julia lts)andQA (julia 1)both failing). It is not introduced here — the PINOODEPDETimeSeriesSolution{...,<:PINOODEMetadata}(p, t)callable (src/pino_ode_solve.jl:379) is ambiguous with the RecursiveArrayTools/SciMLBase(t, ::Type{deriv})interpolation methods. Tracked in QA ambiguities: PINOODE PDETimeSeriesSolution callable ambiguous with RecursiveArrayTools/SciMLBase #1079 and markedaqua_broken = (:ambiguities,)(a tracked@test_brokenplaceholder, not a silent skip). Remove the marker when QA ambiguities: PINOODE PDETimeSeriesSolution callable ambiguous with RecursiveArrayTools/SciMLBase #1079 is fixed.ExplicitImports (all 6 checks)
skip = (NeuralPDE, Base, Core, ModelingToolkit)— the@reexport using ModelingToolkitleaks theModelingToolkitBasemodule name.unwrapwas imported fromSymbolicsbut owned bySymbolicUtils; moved the import toSymbolicUtils(its owner). No ignore needed.get_dvs/get_ivsowned byModelingToolkitBase, accessed via the re-exportedModelingToolkit— ignored (documented).public.LuxCore.{initialparameters,initialstates,parameterlength}are imported to extend their methods (dgm.jl etc.) — dropping them would break method resolution, so they stay imported (non-public in LuxCore).recursive_eltype(Lux) andunwrap(SymbolicUtils) likewise. Ignored, documented.Deps (
test/qa/Project.toml)SciMLTestingcompat -> include"1.6".ExplicitImports(now transitive via SciMLTesting).Aquaas a direct dep (theambiguitiessub-check's child process needs it directly).Verification
Local clean resolution against released SciMLTesting 1.6.0 (no dev-from-branch; resolves SciMLTesting 1.6.0 + Aqua 0.8.16), running the QA group via the run_tests folder model:
0 Fail / 0 Error. The single Broken is the tracked
ambiguitiesplaceholder (#1079).🤖 Generated with Claude Code