On Mon, Apr 26, 2010 at 12:58, Stephen J. Turnbull firstname.lastname@example.org wrote:
It is entirely *not* evident to me that it's too hard to get privileges in the Python development community (Python's development process works -- and it works really well by comparison to 99% of the processes out there).
Well, that's true, all to often a project is controlled by a few developers with no intent of sharing access ever.
Sure, but that's still *work*, and it's work for *somebody else*.
Yes, but only when the checkin was wrong. For all other checkins, it's *less* work. Hence, a committer needs to basically fudge up every second checkin to cause more work than he relieves work. :)
As someone who does a lot more managing of shared resources than coding in the projects I'm active in, I disagree about the danger. Enthusiastic newbies can do a lot of minor damage in a short period of time, and cleaning that up is *work*. This danger is almost entirely mitigated by a small amount of mentoring -- which is precisely what the current process requires -- not only of the recomending party, but also of the existing workers.
And I'm saying that the recommending party is enough. If an established developer says "this guy will not fuck things up", then that is enough guarantee that he won't fuck things up.
But it is a process change, and they should be comfortable with it.
Of course. I'm just arguing for that this process change is correct, and that the current balance is wrong, and that a change is comfortable and safe. :)