On Sat, May 31, 2008 at 11:53:06PM -0400, Karl Fogel wrote:
> While I know Stefan's original mail was well-meant, remember that part
> of the point of the "obvious fix rule" is to remove barries to getting
> obvious problems fixed. Waiting for approval from someone else is a
> barrier -- a small one, often a reasonable one, but a barrier
> nonetheless. So while it's always fine to seek review on a change, a
> commit under the "obvious fix rule" shouldn't also need pre-approval.
> Otherwise, why have an "obvious fix rule"?
Yes. The idea was just that it helps if people make it explicit
when they use the rule, as a way to tell the reviewer "look, I think
this is really simple, don't you agree?".
They don't need to wait for anyone.
Obvious fixes always get reviewed post-commit.
> If on rare occasions we get a failing test because the test output
> wasn't updated, that's an annoyance, but IMHO it's an acceptable price
> to pay for lots of obvious fixes going in with a minimum of overhead.
That is true, and I certainly wasn't trying to be pedantic about
failing tests. This thread was triggered partly by misunderstandings.
I started it because of the discussion about commits made to 1.5.x
directly (still a valid point, I think), the failing test because of
the output change (which Jens had actually fixed in the commit being
discussed, so this was a misunderstanding), the failing windows build
(going back over that thread right now makes it appear much less of
a deal than I made it out to be). The fact that I proposed Jens as a
committer also played a role in my decision to voice my concerns --
after all, I should not be proposing people as committers and then not
keep an eye on their initial commits, right? But by now I think I was
worrying a bit too much over this. The way Jens handled things was great :)
Stefan
- application/pgp-signature attachment: stored
Received on 2008-06-01 10:20:09 CEST