Code review required for commits - formal Team vote.

[Matthieu Patou]

On 10/12/2012 10:48 AM, Jeremy Allison wrote:
> Ok, I'd like to bring this to a conclusion before I have to go
> on my European trip next week :-).
>
> Sorry for asking for another vote after people have already
> +1 and -1'ed, but there have been so many discussions about
> what exactly the details of the policy should be that I think
> it might be useful to re-iterate exactly what we're voting on.
>
> Just to be clear - the formal proposal, originally from
> Ira (but slightly tweaked by me after the discussions)
> is here:
>
> --------------------------------------------------------
> No team member commits their own code.  All code will be "signed off"
> by two team members, as a team member you may sign off your own code.
> The "non-author" team member will be responsible for pushing the code.
> If there are two they can agree among themselves.
>
> Tools for code review to be decided amongst reviewers, but any
> patch posted to samba-technical is an explicit request for review.
>
> Release branches work as they do today, though the + should be treated
> as a sign-off in bugzilla. All release branch bug commits must contain
> their bug number in them, so we can track back what happened.
> --------------------------------------------------------
>
> Given this policy (no author push, no free pass) as an initial
> start I think we can move to a vote.
>
> Yes, there will be things to sort out and discuss as we go
> along, but we're all reasonable (-ish :-) people and can work
> out the minor details to make it work as we go along.
>
> So I know many people have voted back and forth on various
> versions, but I'd like to have a formal Team vote (anyone
> on the Team with a samba.org address) to make a decision on
> this. No reply means abstention.
>
> Let's wait until the end of next week to allow everyone
> who might be travelling or not reading email to respond
> and think carefully about it, and tally the votes on
> Friday 19th Oct (I'll be travelling on that date so
> someone else will have probably to do the count, volunteers
> welcome :-). If it passes let's target implementation on
> Monday 22nd Oct.
>
> I'm voting +1 (in case you wondered).
I like the idea of a more formal code review process but I've expressed 
concerns about the free pass, also I really like that we take the 
approach we did with autobuild.
Which didn't seems to be the approach described in this email (but 
pardon my english).

So it's -1.

Things that will make me change to vote for give it a try:

1) Someone setup a tool for doing the review gerrit, reviewboard, 
tsambareview (yeah we love to have our own tools)
2) Someone checks that it works (ie. he starts to push all his patch for 
review, wait for the review and has the reviewer push his patch)
3) A proposed agenda for testing, like we try for 2 months with the 
following rules but you can still commit directly even if we encourage 
you to use the review process because we tried it (see step 2)

Matthieu.

-- 
Matthieu Patou
Samba Team
http://samba.org