From: Pierre Habouzit Date: Sun, 02 Nov 2008 09:30:30 GMT Subject: Re: [PATCH] prepare deprecation of git-revert Message-ID: <20081102093030.GD4066@artemis> In-Reply-To: <20081102044159.GF5261@coredump.intra.peff.net> On Sun, Nov 02, 2008 at 04:41:59AM +0000, Jeff King wrote: > On Fri, Oct 31, 2008 at 04:55:27PM +0100, Pierre Habouzit wrote: > > > I've not kept the auto-edit feature of git-revert for the git-cherry-pick -R > > case as I don't believe it makes a lot of sense. But if people are unhappy > > with that, I can easily "fix" it. > > I disagree. I write a new commit message for every revert I do. > > When you cherry-pick, you are pulling a good commit from somewhere else. > So its commit message should suffice to explain why you are making the > change (and infrequently, you might want to give more context or say > "and here is where this comes from"). > > But when you revert, you are saying "this other commit was bad, so let's > reverse it." So you can look at the other commit to see what it did, but > you still don't know _why_ it was bad. A revert should always give > information about what you know _now_ that you didn't know when you > made the commit originally. Indeed that makes sense, I'll update the patch then, and be lighter on the deprecation side since it seems I misunderstood what people agreed on. -- ·O· Pierre Habouzit ··O madcoder@debian.org OOO http://www.madism.org