Code review required for commits - formal Team vote.
kai at samba.org
Sat Oct 13 01:33:33 MDT 2012
-----BEGIN PGP SIGNED MESSAGE-----
On 2012-10-13 05:07, Jeremy Allison wrote:
> Well I think we differ on semantics :-). My proposal is a request
> for people to begin the trial. I don't want to try anything new
> without at least a majority of the Team agreeing we should try it,
> which is what I'm asking people to vote +1 on - to try this out.
I don't think this is just a difference of semantics.
> Once we've agreed to try it out, we do that, and if people find
> that patches are getting dropped/not getting addressed in a timely
> fashion, then hopefully someone will propose a change (revert to
> what we had etc. etc) and we'll vote again to get consensus on what
> we need to do about it.
You're basically reverting the burden of proof here. Instead of
proving the new process is an improvement, you're asking people to use
it, and then dissenters need to prove that it's not an improvement.
To over-dramatize an analogy from the legal system, you proposal is
like going from "innocent until proven otherwise" to "guilty until
proven otherwise". You put someone into jail first, and if they're
actually innocent, they're free to start a trial to get out again.
That's not a semantic difference, it's a quite real shift on the
burden of proof.
> The reason I'm big on voting on this is I don't think such a change
> will work without a majority decision to proceed.
If the majority thinks code review is a good thing, why can't they
start using the proposed process without making it mandatory for the
> If it were just trying out a new tool such as autobuild, then yeah,
> we can all do that without needing to formally agree, but if you
> remember we eventually did formally agree on using it. There are no
> tools to test out here, it's an agreement to change the way we're
> doing things.
Yes, but why do we have to buy in before trying the new way of doing
things without committing to it?
> It's been a long time since I worked at a place where code could be
> checked in without a formal review process (I think it was back in
> the 1990's - the EARLY 1990's :-) and in order to produce
> professional software, we really need to adopt professional
> standards of development.
Not all development methods transfer to all sorts of projects. I've
tried extreme programming as proposed by kent beck for a university
project once, and while many of the techniques worked, the pair
programming didn't work out. I figure it didn't work because it
assumes that everybody is stuck in the same room 8 hours a day. That's
not true for university students trying to juggle a project next to
their other coursework.
Basically I'm asking that we prove that the code review process works
before adopting it as a mandatory policy. You're asking us to adopt
it, and then possibly need to prove it doesn't work. I'm afraid it's
going to be hard to prove, because I'm sure the process will work for
Worldforge developer http://www.worldforge.org/
Wine developer http://wiki.winehq.org/KaiBlin
Samba team member http://www.samba.org/samba/team/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://www.enigmail.net/
-----END PGP SIGNATURE-----
More information about the samba-technical