Code review required for commits - formal Team vote.

[Andrew Bartlett]

On Sat, 2012-10-13 at 01:13 +0200, Michael Adam wrote:

> * Productivity need not drop when not pushing ones own
>   patches, because one can work on a private branch until
>   patches hit master.

This assumes there is no work involved in rebasing patches, abandoning
patches that become too hard to rebase, and context switching into areas
that are stuffing from lack of review to beg yet again for review.

I note with interest that nobody proposing this effort is promising to
pick up patches for review, and certainly not in a reasonable
time-frame.  In my particular situation, the only time in the last year
that I've had any significant review available for the AD and
integration changes was when gensec happened to be mutually convenient
for the SMB3 effort.  I appreciated that co-incidence, but I certainly
can't count on it for the rest of my work. 

That the issue of patches being unable to gain review is handwaved off
as 'harass your fellow team members enough and they might actually help
you out' indicates we do not have a mature process here to deal with
this, and we should not jump from our current processes to mandatory
author != committer in just one step. 

Why can't we take it a step at a time, committing for example to have
all substantive changes on the list for 3 days (or until reviewed), but
without rules on author != committer.  This will encourage review
without creating in one step the level of extra work and dislocation
this proposal entails. 

Andrew Bartlett
-- 
Andrew Bartlett                                http://samba.org/~abartlet/
Authentication Developer, Samba Team           http://samba.org