From: Torne Wuff <torne+rockboxdev_at_wolfpuppy.org.uk>
Date: Wed, 7 Sep 2011 11:16:31 +0100

On 7 September 2011 11:09, Torne Wuff <torne+rockboxdev_at_wolfpuppy.org.uk> wrote:
> Anyway. My view on this whole discussion is that I like the idea of
> requiring reviews in principle, but introducing this policy *now*
> would be crazy. We should make the tool available for people to use as
> they want, and see what happens. If people find that reviewing changes
> is a thing that actually happens reasonably often, and a significant
> number of committers choose to have a significant number of their
> changes reviewed voluntarily, then maybe we should discuss requiring
> reviews at that point, but right now we are already changing a number
> of aspects of our workflow because of the tool changes and it's
> already asking quite a lot of everybody to keep up with that. This is
> a pretty mature project with established ways of working, and changing
> too many things at a time without any chance to actually test things
> out is going to cause a lot of unhappiness.

Okay, I sent that kinda prematurely, here's a few more things:

My primary goal in introducing Gerrit was to make it easier for people
*without* commit access to get their changes looked at and submitted,
by reducing the number of steps that a committer has to go through to
do that for them (no mucking about having to patch your local copy
manually with hopefully the right diff downloaded from FS, and so on).
Committers being able to use it to get their changes reviewed is a
nice bonus, but even if nobody ever bothers I still feel the tool is
useful because of its ability to support new contributors.

One thing I would strongly support is requiring review for changes to
release branches (even if that review largely takes the form of "yes I
approve of putting this change into the release branch" rather than
actually being a code review, since most things will probably be
cherry-picks).

-- 
Torne Wuff
torne_at_wolfpuppy.org.uk

Received on 2011-09-07