[openstack-dev] Motion to start using Gerrit for TC votes

Monty Taylor mordred at inaugust.com
Thu Aug 8 13:02:39 UTC 2013



On 08/08/2013 09:30 AM, Thierry Carrez wrote:
> Mark McLoughlin wrote:
>> I think what Monty might be getting at is that TC decisions would no
>> longer happen by default with people voting at a particular TC meeting.
>> Rather a discussion would happen on the mailing list with people casting
>> their votes in gerrit as the discussion evolves and, at some point, a
>> proposal has enough +2 votes (and no -2 votes) such that it's
>> automatically considered passed.
> 
> That still leaves me a bit worried: I don't really want "enough +2" to
> result in passing fast motions, without giving the -2 party a chance to
> convince the others of the flaws in the proposal.
> 
> I guess we can fix that by imposing a few rules on when a vote can be
> considered fully cast (require that the motion is discussed in at least
> one meeting, consider it abandoned/rejected if it doesn't reach enough
> +2 after n meetings).
> 
>> Maybe this could be a good default way for us to decide on things, but
>> if we reach an impasse in gerrit or if a decision has been reached in
>> gerrit by some required date, then we have a final TC IRC meeting and
>> standard meetbot vote to settle it.
> 
> We could still use the review system to settle it during the final
> meeting: just consider non-cast votes as "abstain" and use our quorum rules.
> 
> Something like:
> 
> - Propose motion
> (>4 days -- commenting/voting is open but can't be closed yet)
> - TC meeting about motion
> (7 days -- during that period if the vote gathers enough +2 (or -2) the
> vote will be passed or rejected)
> - Final TC meeting about motion: votes not cast by the end of the
> meeting will be considered "abstain" and quorum rules apply
> 
> (For complex discussions we could have multiple meetings before calling
> for a "final TC meeting".)
> 

I think that's fine for a first stab - but I think we should keep our
eyes on how it goes and see if we can trim it down. I do think what Mark
said is a potential win - as long as we're not losing things. Some
things are quite straight-forward and non-controversial and the
questions people ask could easily be asked on the mailing list or in the
code review. OTOH, some things really do need real-time discussion. I'm
guessing if we do something similar to our current system except only
changing the voting and the fact that a record is produced for a while,
that will be a great step forward. If we feel that mandatory IRC
discussions of every topic are overkill after doing that for a while,
then neat. If not, neat.



More information about the OpenStack-dev mailing list