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

Prerequisites

• Install pixi
• Authenticate with gh
  • If you don't have gh installed, you can pixi global install gh
  • Run gh auth login
  • Run gh auth setup-git
• Clone this repository
  • For example, with gh repo clone rgommers/pixi-dev-scipystack

Setup

In the root directory of this repository:

• run pixi run clone-all to clone all of the source repos into the workspace
  • alternatively, you can select only specific repos to clone, like pixi run clone-numpy
• fork repositories with pixi run fork-all
  • this forks repositories under the username set with gh, and sets upstream and origin remotes appropriately
  • similarly, you can select specific repos to fork, like pixi run fork-numpy

pixi run clone-fork-all will do all of this in one go.

Usage

After that, if you cd REPO (replacing REPO with the name of a repo in the workspace), you will be able to build and test that repo (plus other dev tasks, see pixi task list). For example, with SciPy:

❯ pixi run clone-scipy
{snip}

❯ cd scipy

❯ pixi run build

✨ Pixi task (build in default): spin build --setup-args=-Dblas=blas --setup-args=-Dlapack=lapack --setup-args=-Duse-g77-abi=true  $ meson setup build --prefix=/usr -Dblas=blas -Dlapack=lapack -Duse-g77-abi=true
The Meson build system
Version: 1.8.2
Source dir: /Users/lucascolley/ghq/github.com/rgommers/pixi-dev-scipystack/scipy/scipy
{snip}

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

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

✨ Pixi task (build in default): spin build --setup-args=-Dblas=blas --setup-args=-Dlapack=lapack --setup-args=-Duse-g77-abi=true  $ meson compile -j 8 -C build
{snip}
✨ Pixi task (test-torch in torch): spin test --no-build -b torch -m 'array_api_backends and not slow'
{snip}
======================================== 1945 passed, 14 skipped, 228 deselected in 2.52s =========================================

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!