Skip to main content

Meta-package that provides the oldest NumPy that supports a given Python version and platform. If wheels for the platform became available on PyPI only for a more recent NumPy version, then that NumPy version is specified.

Project description

PyPI

About

This is a meta-package which can be used in pyproject.toml files to automatically provide as a build-time dependency the oldest version of Numpy that supports the given Python version and platform. In case of platforms for which Numpy has prebuilt wheels, the provided version also has a prebuilt Numpy wheel.

The reason to use the oldest available Numpy version as a build-time dependency is because of ABI compatibility. Binaries compiled with old Numpy versions are binary compatible with newer Numpy versions, but not vice versa. This meta-package exists to make dealing with this more convenient, without having to duplicate the same list manually in all packages requiring it.

In other words:

[build-system]
requires = [
    "wheel",
    "setuptools",
    "numpy==1.13.3; python_version=='3.5'",
    "numpy==1.13.3; python_version=='3.6'",
    "numpy==1.14.5; python_version=='3.7'",
    # more numpy requirements...
]

can be replaced by:

[build-system]
requires = ["wheel", "setuptools", "oldest-supported-numpy"]

And as new Python versions are released, the pyproject.toml file does not need to be updated.

Q&A

Why define the Numpy pinnings using install_requires in this repository?

The Numpy version pinnings are defined inside the setup.cfg file as install_requires dependencies, rather than as build-time dependencies inside pyproject.toml. This is deliberate, since Numpy is not actually required to build wheels of oldest-supported-numpy. What we need here is to make sure that when oldest-supported-numpy is installed into the build environment of a package using it, Numpy gets installed too as a runtime dependency inside the build environment.

Another way to think about this is that since we only publish (universal) wheels of oldest-supported-numpy, the wheel contains no pyproject.toml, setup.cfg, or setup.py code - it only contains metadata including dependencies which get installed by pip when oldest-supported-numpy is installed.

Can I use this if my package requires a recent version of Numpy?

In many cases, even though your package may require a version of Numpy that is more recent than the pinned versions here, this is often a runtime requirement, i.e. for running (rather than building) your package. In many cases, unless you use recent features of the Numpy C API, you will still be able to build your package with an older version of Numpy and therefore you will still be able to use oldest-supported-numpy. You can still impose a more recent Numpy requirement in install_requires

What about having a catchier name for this package?

The current name is not very catchy as package names go, but it is very descriptive. This package is only meant to be used in pyproject.toml files for defining build-time dependencies, so it’s more important to have a descriptive than a catchy name!

What if I think that one of the pinnings is wrong or out of date?

Please feel free to open an issue or a pull request if you think something is wrong or could be improved!

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for oldest-supported-numpy, version 0.8
Filename, size File type Python version Upload date Hashes
Filename, size oldest_supported_numpy-0.8-py3-none-any.whl (3.7 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size oldest-supported-numpy-0.8.tar.gz (3.1 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page