Canonical CI: grouped-tests.yml + root test/test_groups.toml#154
Merged
ChrisRackauckas merged 4 commits intoJun 15, 2026
Merged
Conversation
Convert the root test workflow (CI.yml) to the canonical thin caller of
SciML/.github/.github/workflows/grouped-tests.yml@v1, declaring its
group x version matrix once in test/test_groups.toml.
- CI.yml: replace the hand-maintained group x version matrix job with the
reusable grouped-tests.yml caller (coverage:false; on:/triggers preserved
verbatim). Linux-only (no os axis).
- test/test_groups.toml: [Core] on [lts,1,pre] (reproduces the old matrix
exactly) + newly-wired [QA] on [lts,1].
- runtests.jl: add GROUP dispatch (existing suite is Core/All; GROUP==QA runs
the isolated QA env).
- test/qa/{Project.toml,qa.jl}: isolated Aqua + JET environment running
Aqua.test_all(FEniCS) and JET.test_package(FEniCS; target_defined_modules=true).
- Project.toml: raise [compat] julia 1.6 -> 1.10 (LTS floor); add Test = "1"
compat for the test-only extra (benign metadata fixes).
Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ts caller The grouped-tests conversion dropped the Docker container the pre-conversion CI ran in, so every Core job failed with the FEniCS Python stack missing. Pass it through the new container input of grouped-tests.yml. Depends on SciML/.github#86 merging and a v1 retag. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
Author
|
Pushed 5ec5475 restoring the Docker container the pre-conversion CI ran in: the conversion dropped Dependency / merge ordering: this relies on the new 🤖 Generated with Claude Code |
…input The grouped-tests caller passes container: cmhyett/julia-fenics:latest, which was unsupported by SciML/.github grouped-tests.yml@v1 at the time of the last run (startup_failure on 2026-06-10). v1 now resolves to v1.13.0, which adds the container input, so re-run CI to validate. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ts conversion) Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
Author
|
Added the SciMLTesting v1.2 folder-based run_tests harness on top of this grouped-tests conversion; merges as one PR (grouped-tests + SciMLTesting v1.2). |
ChrisRackauckas
added a commit
that referenced
this pull request
Jun 20, 2026
* Fix QA group dispatch: pass QA via the `qa` kwarg, not `groups` The canonicalization conversion (#154) placed the QA group inside the `groups` Dict, but SciMLTesting's `run_tests` routes `GROUP="QA"` through a dedicated `elseif group == "QA"` branch that consumes the `qa` keyword argument and short-circuits before the `groups` table is ever consulted. With `qa` unset, that branch throws ArgumentError: run_tests: GROUP="QA" was requested but no `qa` body was provided so the QA CI job failed on master before loading FEniCS. Moving the QA spec to the `qa` kwarg (matching the documented SciMLTesting form) makes `GROUP="QA"` reach the QA body, while `all = ["Core"]` keeps QA out of the default `GROUP="All"`/local `]test` run as before. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * ci: re-trigger against SciML/.github v1.15.1 Verify the scoped GitHub-hosted runner routing (SciML/.github #97) now sends this repo's apt-packages/container legs to ubuntu-24.04. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Fix FEniCS package-quality (QA) failures surfaced by the QA group The QA group now actually runs FEniCS's own Aqua/JET checks, surfacing six pre-existing package-quality issues. Fix the real causes: - Type piracy + Base.write method ambiguity: `write(path, solution, time::Number)` extended `Base.write` with no FEniCS-owned argument type, which is piracy and also clashed with `Base.write(::IO, x1, xs...)`. Stop importing `write` from Base so `write`/`store` are FEniCS-owned functions (still exported); behavior unchanged. - Undefined export: `export fenicsclass` referenced a nonexistent binding; the macro is `@fenicsclass`. Export the macro. - JET no-matching-method: `Argument(V, number)` lowered to `Argument(::Any, ::Any, ::Nothing)` because `part::StringOrSymbol` cannot accept its own `nothing` default. Widen to `Union{StringOrSymbol, Nothing}`. - Stale dependency: ProgressMeter was declared but unused (only a dead `@require ... using ProgressMeter` no-op). Drop it from deps, compat, and the no-op __init__ block. - deps_compat: add a `[compat]` entry for the LinearAlgebra stdlib. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Keep write/store as Base.write extension; fix piracy+ambiguity via arg types The prior commit removed `write` from the `Base` import to kill the type piracy, but that made `FEniCS.write` a distinct exported binding that clashed with the exported `Base.write` ("both FEniCS and Base export write"), breaking the Core examples (tutorial8/9: `UndefVarError: write`). Restore `import Base: ...write` so `write` stays a single function, and fix the two QA findings on it through argument types instead: - Type piracy: `write` now takes `solution::fenicsobject` (a FEniCS-owned type), so the method is owned by FEniCS rather than pirating Base. The unused, unavoidably-piratical `write(::Any, ::PyObject, ::Number)` overload is dropped (use `store` for raw vectors like `vector(u)`). - Method ambiguity: type `path::PyObject` (what `XDMFFile`/`TimeSeries` return), which is disjoint from `Base.write(::IO, ...)` and `write(::AbstractString, ...)`, removing both ambiguities. `store` is FEniCS-owned (not a Base function), so its overloads are not piracy; `path::PyObject` is added there too for symmetry/correctness. Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: ChrisRackauckas-Claude <accounts@chrisrackauckas.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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
2 participants
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.
Summary
Converts the root test workflow to the canonical SciML thin-caller pattern:
CI.ymlnow callsSciML/.github/.github/workflows/grouped-tests.yml@v1, and the group × version matrix is declared once in a roottest/test_groups.toml.This is a structural conversion. Tests / Aqua / JET were not run locally; they run in CI.
Changes
.github/workflows/CI.yml— replaced the hand-maintainedgroup × versionmatrix job with the reusablegrouped-tests.yml@v1caller.on:/branch triggers preserved verbatim;secrets: inherit.with: coverage: false(this package never collected coverage). Linux-only, so noosaxis. Other workflows left untouched.test/test_groups.toml(new, repo root) —[Core]on["lts","1","pre"](reproduces the old matrix exactly) plus a newly-wired[QA]on["lts","1"].test/runtests.jl— addedGROUPdispatch (default"All"runs the existing suite as Core;GROUP=="QA"runs the isolated QA environment).test/qa/Project.toml+test/qa/qa.jl(new) — isolated Aqua + JET environment runningAqua.test_all(FEniCS)andJET.test_package(FEniCS; target_defined_modules=true), withFEniCSvia[sources]path.Project.toml— benign metadata fixes: raised[compat] juliafrom1.6to1.10(LTS floor); addedTest = "1"compat for theTesttest-only extra (every[extras]dep now has a[compat]entry).Matrix match
compute_affected_sublibraries.jl … --root-matrixemits:The Core rows reproduce the previous
CI.ymlmatrix (group: [Core]×version: [lts, 1, pre], ubuntu-latest) exactly. QA is newly added.Notes
qa.jl.CI.ymlran the Core suite insidecontainer: cmhyett/julia-fenics:latest(providing the FEniCS Python install via PyCall). The reusabletests.ymlhas nocontainer:input, so the canonical caller cannot reproduce that container; onlyapt-packagesis available and no apt package matches the pinned image. The Core runtime FEniCS provisioning therefore needs a follow-up decision (custom runner image / apt-packages / build step) — flagged here for review. This does not affect the QA group (Aqua/JET are static and don't require FEniCS Python).Ignore until reviewed by @ChrisRackauckas.
🤖 Generated with Claude Code