[TYPO3-core] Gerrit and new review policy for TYPO3 v4 Core

[Karsten Dambekalns]

Hi.

On 28.03.2011, at 20:06, François Suter wrote:
> This would seem to mean that a core team member cannot give a +2 unless there's already a +1. He/she would have to give a +1, wait for another +1, convert his/her +1 to a +2 and push. OTOH we could say that we are responsible enough and be allowed to give a +2, while waiting for another +1 to happen, and push only then.

If you give +1 only if you are the first, the second one that reviews can give a +2 if he/she is a team member. If the second one is not a team member and can give only +1, the entry stays in the list without a checkmark, clearly indicating something is missing. Now either the first (team) reviewer goes back and bumps his vote up or a third person picks the change and finishes it. Either way, it works out.

If you give +2 right away, you could have stuck with the existing +1 range - because the major argument was that the list should reflect the "readiness" of a change. If there is a +2, there is a checkmark, although there is only one verified vote. That would render the discussions we had and the change I did useless.

Choose wisely.

> What work process do we want? How exactly does the v5 team do it?

The v5 team did not even think it needs +2 for verified, remember that. :)

I just look at anything open in Gerrit from time to time. If I am about to submit, I see the votes and act accordingly. No strict rules, really, just some mostly unwritten guidelines we discussed some time ago and (probably) put in a daily scrum protocol or something.

> BTW are votes canceled when a new version of a patch is uploaded?

Yes, as the new patch could contain complete new code, a backdoor, whatever. Or simply a new typo that still breaks the code. Not easy for a machine to detect "trivial changes". But there is a feature request open with the Gerrit folks on that topic.

Regards,
Karsten
-- 
Karsten Dambekalns
Gimme Five!
http://typo3.org/gimmefive