Le 09/12/2017 à 08:41, Nick Coghlan a écrit :
On 8 December 2017 at 04:21, Antoine Pitrou antoine@python.org wrote:
Le 07/12/2017 à 19:01, Victor Stinner a écrit :
IMHO the current blocker issue is that it is too hard to become a core developer.
I don't think so. It should not be harder than it was in 2010, yet we are promoting way less core developers than we did. See previous discussion.
I'll note that one thing that *has* changed since then is that we've been putting a lot more emphasis on software distribution via PyPI.
This is sounding a bit like any piece of useful functionality would have been put in the stdlib before 2.7, but that is generally false. Twisted, Numpy, the various Web servers and frameworks out there have never been seriously considered for inclusion AFAIR. There always has been a health ecosystem of third-party libraries that existed happily alongside the main Python distribution.
Conversely, there are some kinds of features that have a much more natural place in the standard distribution rather than third-party libraries. For example object serialization (pickle is a standard, if you have to complement / work around it through third-party hacks it weakens the whole ecosystem) or interfaces to low-level OS functionality (do you want to pip-install sendmsg() or sendfile()?).
I think we've also had a lot of success in bringing the developer experience for core devs and non-core-devs closer together - while *technically* we can push directly to the main repository, very few of us actually do so.
True. But that doesn't change the underlying issue. Core developers are required to actively review and vet proposed changes, and even to maintain important pieces of the stdlib.
Regards
Antoine.