voting in gerrit: merge at +3? (+2+1 / +1+1+1)

Stefan Sperling stsp at
Thu Nov 15 12:15:48 UTC 2018

On Wed, Nov 14, 2018 at 08:44:59PM +0100, Neels Hofmeyr wrote:
> On Wed, Nov 14, 2018 at 04:16:49PM +0100, Stefan Sperling wrote:
> > A transition to "everyone has +2" would be a far more drastic change
> > than what I was proposing. I believe this would result in less code
> > review being done than is being done today.
> In general, the fact that you can do +2 doesn't mean that you're allowed to do
> it. The normal case would be: the third reviewer coming along sees that there
> are already two +1s and then is allowed to vote a +2, by convention / trust.
> I would like to adopt the +3 scheme, only I think we don't necessarily need IU
> enforcement of it by an obscure plugin, because we can agree on it between us
> humans. Whenever someone disregards "+3", that is a potential offense.
> I could trivially enable +2 within minutes right now.  No prolog required, only
> communication.  If that doesn't work out, we can still go for the plugin or
> revert +2 abilities. What do you think?
> ~N

I believe that if we're going to use tooling to support a process,
then our tooling should be configured to support the process we want.
If our tooling is not or cannot be configured as such, we might as well
use different tooling. 

Social conventions will work but only with continuous effort. Try to think
of this as somebody who is looking at Osmocom from the outside without
any prior knowledge or involvement, and tries to figure out how we work.
The easier it is for that person to figure out answers, the better.

If gerrit does something else than our process says it should,
then gerrit makes things harder for new contributors, not easier.

More information about the OpenBSC mailing list