Le 13/07/2018 à 13:31, Nathaniel Smith a écrit :
I volunteer to co-author such a PEP. But I'm not up to doing it on my own. So... who else wants to be a co-author? (I'm not going to pressure anyone, but Brett, Mariatta, and Carol, please know that your names were the first ones that jumped to my mind when thinking about this :-).)
I don't know how much time I'll be able to devote to it, but feel free to enlist me.
If you have suggestions for particularly interesting projects or excellent writing on the topic, then this thread would be a good place to mention them.
Perhaps Apache httpd? (or some other major Apache project, since I /think/ they share similar governance structures... I happen to work on Apache Arrow, which is young and a bit on the small side compared to Python, but can ask the project leaders for feedback)
What I'm thinking:
- While this might eventually produce some recommendations, the
immediate goal would just be to collect together different options and ideas and point out their trade-offs. I'm guessing most core devs aren't interested in becoming experts on open-source governance, so the goal here would be to help the broader community get up to speed and have a more informed discussion .
- As per the general PEP philosophy, I think this is best done by
having some amount of general discussion on python-dev/python-committers, plus a small group of coauthors (say 2-4 people) who take responsibility for filtering ideas and organizing them in a coherent document.
- Places where we'll want to look for ideas:
- The thread already happening on python-committers
- Whatever books / articles / blog posts / etc. we can find (e.g. I
know Karl Fogel's Producing OSS book has some good discussion)
- Other major projects in a similar position to CPython (e.g.,
node.js, Rust) -- what do they do, and what parts are they happy/not-happy about?
- Large Python projects (e.g. Django) -- likewise
 The NumPy project has put a lot of energy into working through governance issues over the last few years, and one thing that definitely helped was coming up with some "assigned reading" ahead of the main sprint where we talked about this. NumPy's problems are/were pretty different from CPython's, but I'm imagining this PEP as filling a similar role.