Re: Commitfest problems

Josh Berkus <josh(at)agliodbs(dot)com> writes:
> Well, the CF process was added for 2 reasons:

> 1) We were losing track of some patches until integration time, when
> Bruce suddently found them in the -hackers archives.

> 2) In order to give the senior committers some time *off* from reviewing
> other people's patches. The idea was to spend 3 weeks or so intensively
> reviewing others' patches, and then to have a month (or more) *off* from
> working on anything but your own stuff.

Right; I was in the middle of composing words to the same effect when
this arrived.

> While the CFs are still doing (1), support for (2) ended sometime in the
> 9.3 development cycle. Partly this is because current CFMs are not
> permitted to take authoritative steps to ensure that the CF ends on
> time, and partly it's because of the increase in big complicated patches
> which just don't fit into the CF cycle.

I don't see why you think CFMs are not "permitted" to close out a CF
when they want to. At least some of the fests have been closed out per
calendar schedule, punting unprocessed patches to the next fest. We've
certainly utterly failed to do that since August, but I think that's
mismanagement.

> Speaking as the originator of commitfests, they were *always* intended
> to be a temporary measure, a step on the way to something else like
> continuous integration.

Really? How would that address (2)? And please note that part of the
problem we're having right now is that senior people are rebelling
against no longer having any agreed-on time off. IMV, goal (2) is
not optional; if you try to force people into continuous review work,
you'll just lose them completely.

regards, tom lane