From: Jeff King <peff@peff.net>
Date: Fri, 22 Nov 2013 22:32:22 GMT
Subject: Re: What's cooking in git.git (Nov 2013, #05; Thu, 21)
Message-ID: <20131122223222.GA7538@sigill.intra.peff.net>
In-Reply-To: <xmqqr4a8yzga.fsf@gitster.dls.corp.google.com>

On Fri, Nov 22, 2013 at 12:05:25PM -0800, Junio C Hamano wrote:

> > If this could make it to `next` some time next week, that would work
> > out great for us, because we may start considering using `next` as a
> > partial or full deployment on our production machines
> 
> I do not think potentially incompatible stuff that are slated for
> 2.0 that have been cooking in 'next' affects the server side, so
> that may be a good and safe move.

I do not think we will literally run `next` in this case, but probably
v1.8.5 + selected topics (like this one :) ).

We do not need to base ourselves on a release, of course, and we may
start using a rolling version of master, but choose quiescent points in
the cycle (like starting with a release, and then rolling forward around
-rc time). I started trying that with this cycle, which is how I found
the --literal-pathspec regression in mid-cycle, and then found out the
fix hadn't graduated during -rc. :)

If that proves stable, then I will consider bumping up our frequency of
following `master`, and then eventually following `next` (possibly with
some lag). As a large site, we get to expose the code to a lot of new
people; but we also need to be mindful that we are exposing a lot of
people to new bugs.

-Peff