pixi-dev-scipystack

This is an experimental development setup for using pixi to do development work on NumPy, SciPy and some related projects.

Getting started

Install pixi, clone this repo, and then for the project you want to work on ensure that you have a clone of the upstream repo one level below the subdirectory for the project containing pixi.toml. For example:

% git clone https://github.com/rgommers/pixi-dev-scipystack.git
% cd scipy
% git clone https://github.com/scipy/scipy.git
% (cd scipy && git submodule update --init --recursive)

After that, you will be able to build and test SciPy (plus other dev tasks, see pixi task ls):

% pixi run build
✨ Pixi task (build in default): python dev.py build
💻  meson setup /Users/rgommers/code/pixi-dev/scipy/scipy/build --prefix /Users/rgommers/code/pixi-dev/scipy/scipy/build-install
The Meson build system
Version: 1.5.1
Source dir: /Users/rgommers/code/pixi-dev/scipy/scipy
...


% pixi run test-torch
? The task 'test-torch' can be run in multiple environments.

Please select an environment to run the task in: ›
❯ pytorch
  array-api

✨ Pixi task (test-torch in pytorch): python dev.py test -b pytorch -s cluster
💻  ninja -C /Users/rgommers/code/pixi-dev/scipy/scipy/build -j8
ninja: Entering directory `/Users/rgommers/code/pixi-dev/scipy/scipy/build'
[2/2] Generating scipy/generate-version with a custom command
Build OK
💻  meson install -C build --only-changed --tags runtime,python-runtime,tests,devel
Installing, see meson-install.log...
Installation OK
SciPy from development installed path at: /Users/rgommers/code/pixi-dev/scipy/scipy/build-install/lib/python3.12/site-packages
Running tests for scipy version:1.15.0.dev0+1346.64a49b0, installed at:/Users/rgommers/code/pixi-dev/scipy/scipy/build-install/lib/python3.12/site-packages/scipy
...
======================================================= 125 passed, 23 skipped in 2.46s =======================================================

Goals

Goals of this repository include:

1. Making it easier to get started contributing to projects like NumPy and SciPy.
2. Making it easier for maintainers to reliably test in non-standard environments, e.g. when working on support for PyTorch, JAX, CuPy and other array libraries, or with BLAS/LAPACK libraries like MKL.
3. Prototype extra spin/dev.py commands that may be useful.
4. See what else Pixi can bring!

Contributing

This repo is highly experimental - everything is subject to change, so please don't rely on it too much. Ideas, questions, bug reports and pull requests are very welcome though!