Ian Hickson wrote:
>
> On Wed, 30 May 2001, Mitchell Baker wrote:
> >
> > http://www.mozilla.org/hacking/code-review-faq.html
>
> A very helpful document!
>
> I have one comment (stolen from hyatt): What if a super-reviewer sees a
> patch is flawed, but the patch manages to get r= and sr= through different
> people? Ideally this would never happen, but it has. How can
> super-reviewers block a check-in? Is there an equivalent of "veto="?
>
> How about other people? Is there a way for me, for instance, to block a
> check-in based on a fundamental flaw in the patch, even if it passes r=
> and sr= from different people?
>
> Assuming the answer to both is 'no', I would like to propose that super-
> reviewers be able to say veto=jsmith. (Other people (like me) would then
> be able to ask super-reviewers to veto a patch. I don't think there is any
> need for anyone to be able to block check-ins without going through a
> respected community member first.)
Flaws are demonstrable. Point one out and people will see it.
Doing so in english is good enough. This was discussed the other
day and we agreed that process was not required. If a hacker gets
an r= and sr= and then someone points out a flaw and the hacker
checks it in anyway (while aware of the previously unnoticed
flaw) then that hacker is in the wrong and this will be clear to
all.
Differing opinions about whether or not a given change is the
best change is harder. Discussion is still the answer. Sometimes
you win, sometimes you lose.
>
> I also have one minor question. What's the difference between "rs=jsmith"
> and "sr=jsmith"? A rubber-stamp still needs a quick look at the code, to
> decide if its small enough for being rubber-stamped, right? Or is a rubber
> stamp actually a review, with the statement that no super-review is
> required? (i.e., is 'rs=x' the equivalent of 'sr=x' or is it the
> equivalent of 'r=x, sr=n/a'?) I think the FAQ could be slightly more
> explicit about this.
rs= makes the claim that that super reviewer has briefly glanced
at the change and trusts that it will be OK but does not believe
there is a need to super review it closely. I'd argue that an r=
should still be required. But I don't know the party line on that
issue. I remain of the opinion that hackers *should* get review
before even asking for sr, but that is not a rule per se.
John.
>
> --
> Ian Hickson )\ _. - ._.) fL
> Netscape, Standards Compliance QA /. `- ' ( `--'
> +1 650 937 6593 `- , ) - > ) \
> irc.mozilla.org:Hixie _________________________ (.' \) (.' -' __________
