On 04/17/2012 02:53 AM, Kevin Wolf wrote:
Am 16.04.2012 23:47, schrieb Anthony Liguori:
On 04/16/2012 04:24 PM, Peter Maydell wrote:
I don't particularly object to providing a T: line for
target-arm.next/arm-devs.next, but I'm not sure it's particularly useful,
since we don't have the same tendency the kernel does to having subtrees
which can diverge significantly because of large amounts of change waiting
for a merge window. I wouldn't expect people to base arm patches against
arm-devs.next rather than master, for instance. (Maybe I should??)
Anyway, I think if we have T: lines in MAINTAINERS it should be because
(and we should clearly say that) that is the tree that we expect patches
in that area to apply to.
I think we should (and do already?) say that all patches on qemu-devel should be
against qemu.git unless otherwise indicated in the patch subject.
In very busy phases I have asked people to base their patches on the
block branch in order to avoid conflicts and save unnecessary work for
both the author and the maintainer.
Yeah, I think this is really the exception that proves the rule.
A patch based on master that conflicts with the subtree is useless, it
must be rebased anyway. So the theory is that you base patches sent to
me on the block branch rather than master. In practice, the difference
between qemu subtrees is so small that in most cases basing directly on
master works just fine, but it's not the safe way and I may ask you to
rebase.
I think the block subtree is really an exception simply because it's the oldest
and best established subtree. I think requesting patches be against it is
reasonable.
It's up to you as to what you want the standard policy to be. If you want
patches against the block tree normally, I'd just request having folks add a tag
like they do for the qemu-kvm tree ([uq/master]).
Regards,
Anthony Liguori
Kevin