task: agent ready prep by jharlow-intel · Pull Request #280 · IntelPython/mkl_umath

jharlow-intel · 2026-10-07T18:47:02Z

Very similar to IntelPython/mkl_fft#379: adds contributor and coding-agent guidance (CONTRIBUTING.md, AGENTS.md), PR and issue templates, and a README pointer, checked against this repo's build and CI. It also corrects the existing AGENTS.md files.

- CONTRIBUTING.md: use --override-channels, install pre-commit with the pylint and shellcheck its system hooks need, note Windows support and the meson.build GCC/MSVC branches, and match the CHANGELOG link format - AGENTS.md: restore supported platforms, add the meson.build guardrail, and note that checkout tests need the editable install - conda-recipe-cf/AGENTS.md: gnu_thread is Linux-only, llvm-openmp has no run pin, and each recipe has its own conda_build_config.yaml - mkl_umath/src/AGENTS.md: scope the patch lock claim and note per-thread patch/restore balance - .github/AGENTS.md: list the templates and repo config files

napetrov

I checked the docs at 3f8573f against the tree and by running them.

I ran the CONTRIBUTING.md setup verbatim in a clean condaforge/miniforge3 container. Creating the env, the editable install, pytest mkl_umath/tests (392 passed, 1 skipped) and pre-commit run --all-files all pass. A control build without -Dmkl_threading=gnu_thread fails at import with libmkl_intel_thread.so.3: undefined symbol: __atomic_compare_exchange, so the gnu_thread note is right.
I also ran a static check of the paths, commands and identifiers the guidance names. At head it finds no broken references. At base it found the stale mkl_random mentions and the src/__umath_generated.c path, which this PR fixes.
I checked loading with Claude Code 2.1.296. The root AGENTS.md loads from every directory, and each nested one loads once the agent works in its subtree. .github/copilot-instructions.md is never loaded; more on that inline.

A few remaining points are inline. None of them block the PR.

napetrov · 2026-10-13T22:09:49Z

+On Linux:
+
+```sh
+conda create -n mkl_umath-dev --override-channels -c conda-forge python pip \


python is unpinned here. conda-forge already has Python 3.15.0, while pyproject.toml has requires-python = ">=3.10,<3.15". Today the solve picks 3.14.8 (checked), likely because not every dependency has a 3.15 build yet. Once they all do, this line creates a 3.15 env and the editable install fails on requires-python. I suggest "python<3.15", or a specific python=3.14.

Python 3.15 is going to be enabled in #279

napetrov · 2026-10-13T22:09:49Z

+- Treat recipe files as canonical for packaging intent and dependency pins.
+- Keep recipe changes in step with `.github/workflows/conda-package.yml`, which
+  builds this recipe and runs the full test suite against the result.
+- The package test in `meta.yaml` runs only `test_basic.py`.


It's worth saying that CI never runs this package test. Both conda workflows call conda build --no-test (conda-package.yml:286, conda-package-cf.yml:275) and run pytest --pyargs mkl_umath in a separate job. Without that note, an agent can assume CI covers a change to the meta.yaml test.

Just to note, we run that through the Jenkins currently

napetrov · 2026-10-13T22:09:49Z

+## Development guardrails
+- Preserve NumPy ufunc behavior; patched loops stand in for NumPy's own.
+- Edit the `*.src` templates and `generate_umath.py`, not generated C.
+- Keep the floating-point precision flags in `meson.build`; Intel-only flags stay


/fp:precise is also set for Windows in both recipes' conda_build_config.yaml (line 20, # [win]), not only in meson.build. Could this line, and CONTRIBUTING.md:47, say "in meson.build and the recipes' conda_build_config.yaml"? That way an agent editing one place knows there is a second.

napetrov · 2026-10-13T22:09:49Z

 ## Contribution expectations
 - Keep changes atomic and single-purpose.
- Preserve runtime patching API (`use_in_numpy()`, `restore()`, `is_patched()`) unless explicitly requested.
+- Preserve the runtime patching API (`patch_numpy_umath()`, `restore_numpy_umath()`,


This rule exists only in copilot-instructions.md. Claude Code never loads that file (checked with 2.1.296 from every subdirectory), and agents other than Copilot see it only if they follow the pointer at AGENTS.md:53. If the rule should apply to all agents, copying this line into the root AGENTS.md guardrails would cover them.

Also, use_in_numpy() and restore() are still exported in __all__ and exercised in test_patching.py. If they should stay until the deprecation removal, list them here too, or say that they may be removed.

antonwolfy · 2026-10-14T11:31:21Z

+- `build_pip.yml` — editable pip build with `icx`, including pre-release NumPy
+- `build-with-clang.yml` — build with `icx` from the oneAPI apt repository
+- `build-with-standard-clang.yml` — build with upstream clang


The three pip/clang workflows also run pytest --pyargs, but the doc says "build" only.

antonwolfy · 2026-10-14T11:33:40Z

+      description: |
+        Output of:
+        ```
+        python -c "import mkl_umath, numpy; print(mkl_umath.__version__); print(numpy.__version__)"


It never asks for the MKL version. Suggest conda list mkl / pip list | grep -i mkl; mkl-service isn't a dependency here.

antonwolfy · 2026-10-14T11:35:06Z

+body:
+  - type: markdown
+    attributes:
+      value: |


With |, GitHub keeps each newline as a hard line break when rendering the form, so these wrapped lines break mid-sentence. The rendered form at this commit has follow<br>SECURITY.md. Same at lines 59-62 and in feature_request.yml:17-19.

Putting each paragraph on one line fixes it. Keep | for the Versions description, which needs it for the code fence.

antonwolfy · 2026-10-14T11:35:57Z

+```
+
+`--override-channels` keeps MKL and its OpenMP runtime from mixing with your
+configured channels. `gnu_thread` is required with conda-forge's MKL; the


Say "On Linux, gnu_thread is required…", and add that Windows keeps the intel_thread default.

antonwolfy · 2026-10-14T11:40:32Z

+
+## Pull requests
+
+Work on a branch and fill in the PR template. For user-visible changes, add a


Put "Work on a branch" before the Checks section.
On a branch main, the no-commit-to-branch hook makes pre-commit run --all-files exit 1; on any other branch it passes.

antonwolfy · 2026-10-14T11:43:35Z

- No `conda_build_config.yaml` (uses conda-forge defaults)
- May use different compiler toolchain
- For conda-forge feedstock integration (if upstreamed)
+## Differences from `conda-recipe/`


Add the NumPy run pin, since cf uses pin_compatible('numpy')

antonwolfy · 2026-10-14T11:44:04Z

+- Keep inputs deterministic; benchmarks seed their own RNG.
+- Report performance numbers with reproducible context: hardware, thread count,
+  versions, and the command used.
+- Results under `benchmarks/.asv/` are local artifacts.


Write .asv/, not benchmarks/.asv/. That matches the file's own dir-relative paths.

antonwolfy · 2026-10-14T11:46:38Z

 pip-log.txt
 pip-delete-this-directory.txt

+# Virtual environments, test caches, local env files


Do we need to .gitattributes with *.sh eol=lf, matching mkl_fft?

antonwolfy · 2026-10-14T11:48:41Z

+- Prefer updating upstream source when feasible; keep local vendored diffs
+  minimal.
+- Do not refactor vendored code opportunistically in unrelated PRs.
+- `black` and `isort` are configured to skip the vendored files


flake8 also skips these files (exclude in .flake8). codespell, pylint and the whitespace/EOF hooks still run on them.

antonwolfy · 2026-10-14T11:49:58Z

- Patching: `patch_numpy_umath()`, `restore_numpy_umath()`, `is_patched()` state transitions
- Edge cases: NaN, Inf, empty arrays, large arrays
- Dtype coverage: float32, float64, complex64, complex128
+## Expectations


Worth adding to Expectations, that the tests that patch must restore before returning, in the same thread (try/finally or with mkl_umath.mkl_umath():).

jharlow-intel requested review from antonwolfy, ndgrigorian, vlad-perevezentsev and xaleryb as code owners October 7, 2026 18:47

task: agent ready prep

823e526

jharlow-intel force-pushed the task/agent-ready branch from c80b57b to 823e526 Compare October 7, 2026 18:52

task: clean up CONTRIBUTING.md

9cefba6

antonwolfy reviewed Oct 13, 2026

View reviewed changes

jharlow-intel requested a review from antonwolfy October 13, 2026 19:00

napetrov reviewed Oct 13, 2026

View reviewed changes

antonwolfy reviewed Oct 14, 2026

View reviewed changes


		## Pull requests

		Work on a branch and fill in the PR template. For user-visible changes, add a

Conversation

jharlow-intel commented Oct 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

napetrov left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

jharlow-intel commented Oct 7, 2026 •

edited

Loading