Thanks for bringing this up. We’ve never actually had that discussion, so the official policy is very loose.
The practice is for committers to use their discretion. No committer needs to ask for permission, but it is polite and prudent to ask for review if it is a large change and/or in an area that the committer is not familiar with. In my opinion, the current practice seems to be working well. People are not subjected to undue delays waiting for review, we are not turning away would-be contributors by being too harsh, nor are we admitting dubious code into the code base. But I would say that! It is a valid criticism that Calcite is somewhat of a benevolent dictatorship, with me as the dictator, writing a lot of the code and doing a lot of the reviews. Other committers tend to ask for their commits to be reviewed, perhaps out of deference. Benevolent dictatorships work well in other open source projects (think Linux[1], Python, Ruby), but are not the Apache Way. I want to dismantle the perception and reality that the project is run that way. If a more explicit and prescriptive policy would help make Calcite more egalitarian, let’s consider it. Julian [1] http://lkml.iu.edu/hypermail/linux/kernel/1510.3/02866.html <http://lkml.iu.edu/hypermail/linux/kernel/1510.3/02866.html> > On Nov 11, 2015, at 2:33 PM, Josh Elser <[email protected]> wrote: > > Hi, > > I'm being lazy and not looking through history, but has there been any > discussion/consensus on the general topic of > commit-then-review/review-then-commit for Calcite? It seems like most of the > time it's just discretionary, but I figured I should ask instead of silently > assuming. > > Thanks!
