Re: [Libreoffice] [META] nitpicking on cherry-picking review rules

Lionel Elie Mamane píše v Po 22. 08. 2011 v 18:01 +0200:
> On Mon, Aug 22, 2011 at 02:24:09PM +0200, Jan Holesovsky wrote:
> > On 2011-08-21 at 08:36 +0200, Lionel Elie Mamane wrote:
>
> > > I'm unsure whether the acks needed for cherrypick are "any committer"
> > > or a more restricted group, but in case it is the former: I agree that
> > > commit 03e9161e2eca9d389d7ce419495538c31f6aed31 is worthwhile to
> > > cherrypick for 3.4.3, as:
>
> > Any developer counts, so you are welcome to review patches!  Only in
> > this case, you are the author of the patch - a self-review does not
> > count ;-)
>
> I did not notice that Noel already counted his review as part of the
> three needed, I had (wrongly) in mind that three *more* than Noel were
> expected. I thought that if Noel does not count, I ought to; in other
> words, it should not be the case that both the author and the person
> asking for cherry-pick be excluded in the count.
>
> Sorry for the mix-up in that; if Noel counts as one of the three, it
> is reasonable that I don't, although if I had to decide on the rule,
> I'd prefer "need three reviews (and author counts) PLUS the person
> asking for cherry-pick", because this is more regular in the face of

I think that the confusion is that it is not clear who is author and who
approved the fix in the 3-4 branch. I guess that Noel kept you as the
author and used himself as the reviewer.

Person asking for cherry picking does not count. It can be Johny Tester
who feels that this fix is important enough for the bugfix release. Only
people who really review the code counts :-)

Regarding this particular commit. You need 1 approval for 3-4 branch and
two more approvals for 3-4-3 branch. Noel already committed it into 3-4
branch, so you need two more approvals for the 3-4-3 branch. It is clear
that it can be neither Noel nor you. One of you was author and the other
was reviewer for the 3-4 branch.

> the situation where the author does not agree to the cherry-pick; the
> current rule basically ignores the author's "not agree", while my
> amended rule expects him/her to be replaced by another developer.

If the author does not agree with cherry picking, he could simply
complain in the mail. I am sure that the fix will not be pushed until
all complains are clarified.

Anyway, thanks for the fix. I hope that the processes are more clear now.


Best Regards,
Petr