[SCM] Samba Shared Repository - branch master updated

[Michael Adam]

Hi Volker,

Jeremy Allison wrote:
> On Fri, May 18, 2012 at 08:51:51PM +0200, Volker Lendecke wrote:
> > On Fri, May 18, 2012 at 11:48:35AM -0700, Jeremy Allison wrote:
> > > On Fri, May 18, 2012 at 08:44:34PM +0200, Volker Lendecke wrote:
> > > > On Fri, May 18, 2012 at 11:40:07AM -0700, Jeremy Allison wrote:
> > > > > On Fri, May 18, 2012 at 08:37:12PM +0200, Volker Lendecke wrote:
> > > > > 
> > > > > > Andrew Bartlett was blocked from working. Yes, I might do
> > > > > > reverts in a hurry, but it was the only possible option I
> > > > > > saw in the time frame I had available. I just did not have
> > > > > > time to fix the root cause of the flakey builds.
> > > > > 
> > > > > You have no evidence that your patches were the cause
> > > > > of the flakey builds. In this situation reversion was
> > > > > unwise.
> > > > 
> > > > I had Andrews firm suspicion together with the inability to
> > > > fix it. This is enough for me to pull it.
> > > 
> > > Well I'm asking you not to be so arbitrary in the future
> > > please. It really doesn't help.
> > 
> > Metze being fully aware, Obnox having at least heard about
> > it, Andrew B. suspecting problems, do you really call that
> > arbitrary?

You can't count me as a lever on that. I had heard that you asked
Metze to cancel his autobuild because you had an autobuild running
that was conflicting in some sense. I also sensed some strange
urgency, but I didn't get what this was about at that moment.

Also, you are trying to justify retrospectively that you did
actually have implicit ACK for the revert patch.

If I had been aware of the issue, I would have argued precisely
as Jeremy, i.e. that you should not revert the changes just based
on a suspicion that they may have caused some flakeynes.

> Yes I do. The justifications here are ones you've added
> after the fact (I've watched you do it :-). Your earliest
> reaction of "this is MY code I can do what I want with it"
> was the most honest.
> 
> The fact is you felt your code was being criticised, and
> withdrew it without consultation.
> 
> Please don't do that.

I think this analyis of Jeremy's is minute.

Even legitimate criticism is no reason to just revert arbitrarily.
Instead, one should try to improve the situation once
the patches have hit upstream (whether by the author's own
push or by a reviewer). I also share the point of view that you
loose the exclusive ownership of the patches once the hit
upstream and hence the exclusive right to arbitrarily revert them.

A revert should only be the last resort, and I agree that we
should set up the policy of allowing revert only with explicit
ACK (i.e sign-off).  Imho, it makes perfect sense to have this
rule for the special case of reverts to reduce arbitrary acts.

One more thought: Your patches may have been pushed prematurely.
(But they have been reviewed by Jeremy and had passed autobuild
at least once. So they were not completely wrong for sure.)
But for the current argument it is important that it was not you
who pushed the patches. So you need not feel put under pressure
on your free days if they are under suspicion cause any grief.
This can be no argument for you do push a revert in a
panic-action. There was no reason to panic-act. And if at all
then it would have been Jeremy's part to trigger that.

Cheers - Michael

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 206 bytes
Desc: not available
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20120519/88b8b9af/attachment.pgp>