Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Prevent package from being installed on old Python versions

What can we put in a setup.py file to prevent pip from collecting and attempting to install a package when using an unsupported Python version?

For example magicstack is a project listed with the trove classifier:

Programming Language :: Python :: 3 :: Only

So I expect the following behaviour if pip --version is tied to python 2.7:

$ pip install magicstack
Collecting magicstack
  Could not find a version that satisfies the requirement magicstack (from versions: )
No matching distribution found for magicstack

But the actual behavior is that pip collects a release, downloads it, attempts to install it, and fails. There are other Python3-only releases, curio for example, which actually install fine - because the setup.py didn't use anything Python 3 specific - only to fail at import time when some Python 3 only syntax is used. And I'm sure there are packages which install OK, import OK, and maybe only fail at runtime!

What is the correct method to specify your supported Python versions in a way that pip will respect? I've found a workaround, involving uploading only a wheel file, and refusing to uploading a .tar.gz distribution, but I would be interested to know the correct fix.


Edit: How does pip know not to download the wheel distribution if the Python/os/architecture is not matching? Does it just use the .whl filename convention or is there something more sophisticated than that happening behind the scenes? Can we somehow give the metadata to a source distribution to make pip do the right thing with .tar.gz uploads?

like image 373
wim Avatar asked Feb 15 '17 00:02

wim


People also ask

Can I install two versions of a Python package on the same computer?

With maven or gradle you can install two versions of the same package, with pip you cant. Still you cant use two versions of the same package in the same program.

Does pip automatically install latest version?

Using Requirements Files. The pip install command always installs the latest published version of a package, but sometimes your code requires a specific package version to work correctly.

How do I install Python packages to a specific version?

To install a specific version of a Python package you can use pip: pip install YourPackage==YourVersion . For example, if you want to install an older version of Pandas you can do as follows: pip install pandas==1.1.


1 Answers

There is a correct way to do this, but unfortunately pip only started supporting it in version 9.0.0 (released 2016-11-02), and so users with older versions of pip will continue to download packages willy-nilly regardless of what Python version they're for.

In your setup.py file, pass setup() a python_requires argument that lists your package's supported Python versions as a PEP 440 version specifier. For example, if your package is for Python 3+ only, write:

setup(
    ...
    python_requires='>=3',
    ...
)

If your package is for Python 3.3 and up but you're not willing to commit to Python 4 support yet, write:

setup(
    ...
    python_requires='~=3.3',
    ...
)

If your package is for Python 2.6, 2.7, and all versions of Python 3 starting with 3.3, write:

setup(
    ...
    python_requires='>=2.6, !=3.0.*, !=3.1.*, !=3.2.*, <4',
    ...
)

And so on.

Once you've done that, you will need to upgrade your version of setuptools to at least 24.2.0 in order for the python_requires argument to be processed; earlier versions will just ignore it with a warning. All of your project's sdists and wheels built afterwards will then contain the relevant metadata that tells PyPI to tell pip what Python versions they're for.

like image 157
jwodder Avatar answered Nov 16 '22 02:11

jwodder