/hg/release/icedtea6-1.8: PR632: 6878713.patch breaks shark zero...
Dr Andrew John Hughes
ahughes at redhat.com
Fri Mar 25 17:16:27 PDT 2011
On 17:00 Fri 25 Mar , Mark Wielaard wrote:
> On Fri, 2011-03-25 at 15:34 +0000, Dr Andrew John Hughes wrote:
> > > I just tried to be helpful when Matthias asked if I thought it was a
> > > good idea to backport that patch. And I thought it was, even though I
> > > don't use that branch myself, but obviously Matthias does.
> > >
> > I don't have a problem with that. But there's a missing stage between you
> > saying "that sounds a good idea" to a backported patch appearing on the branch
> > with no review. The actual patch for 1.8 wasn't posted before commit, and there's
> > no indication as to what builds have been tested, if any.
> I think this is where the disagreement it. If Matthias asks, then I
> trust him to only commit if it applies cleanly and he has tested it.
> That should be obvious. Nobody should commit something they didn't test
> themselves. The patch seemed trivially correct in the context to me. But
> I didn't commit it myself since I don't test the branch, which is why I
> asked Matthias to do it, because he does.
> Yes, it is a judgment call. If the patch was larger than 3 lines, there
> might be some more discussion. If it was someone else than Matthias,
> there might have been more reservation of just approving. But in general
> if I ack a patch I trust people to do the right thing wrt making sure
> patches apply cleanly and the result is tested.
I try to treat everyone equally and not be more lax just because someone
has been committing to the project for a longer period of time. I also
don't see it as an issue of trust, but a simple acknowledgement that everyone
makes mistakes sometimes and it's better to double-check every patch for
a release branch. This is also reinforced by the fact that patches aren't
reviewed for HEAD, so they end up getting applied, sometimes without any
explanation of why (e.g. the other fix posted to 1.8).
I don't see what the size of the patch has to do with anything so please
stop throwing that in to confuse things. A three-line patch can break
a build while a hundred line patch doesn't.
Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)
Support Free Java!
Contribute to GNU Classpath and IcedTea
PGP Key: F5862A37 (https://keys.indymedia.org/)
Fingerprint = EA30 D855 D50F 90CD F54D 0698 0713 C3ED F586 2A37
More information about the distro-pkg-dev