Given that the MMAP backed array library I have been working on is now in use with multiple projects I figure that the time has come to put it up on PyPi. The project started off as an in-house library that was used in a few projects, given we were making something fast for our internal use we didn’t spend a lot of time on making a nicely packaged library. Being able to have an editable pip install from a git repository was good enough. It’s been extremely useful for me so it is probably useful for other people too, the time had come to do some housekeeping and get it out there on PyPi.
That of course means doing all the things on the Python packaging user guide. Many of those things were already done but I needed to do the following:
- Add a license file
- Put the license type into the setup.py under license
- Put my name in as the author in setup.py under author
- Create a PyPi account and upload the package
- Put a PyPi badge in the README file 😉
Then after that I ran it all through cheesecake to make sure I hadn’t missed anything critical
However these are rather small technical details that I can just grind through in an afternoon. They are necessary but not sufficient conditions for making a healthy open source library. The bigger issue is…
Making it easier for people to contribute
The bigger issue I wanted to address is making the project more friendly to contributors. I just finished watching VM Brasseur’s fantastic PyConAU 2016 talk about encouraging drive-through contributors and I realize there’s significant improvements that can be made with this library.
Specifically with this library:
- I needed to put some effort into describing the contribution process. That meant writing a CONTRIBUTORS file that explains how to make contributions. In this I tried to make it clear how to make pull requests and what’s required to get them merged in. This resulted in starting to put together some python coding guidelines so that new contributors can get up to speed quickly.
- Explain how people can get involved in the project.
- Host documentation on readthedocs or similar.
If there’s anything else you can see that would make it easier to contribute I’d love to know, please leave a comment if there’s anything I’ve missed.
The mmap_backed_array package is now up on PyPi!