Re: git bug/feature request

From: Daniel Barkalow <barkalow@iabervon.org>
Date: Nov 27, 2007, 20:19 UTC
Message-ID: <Pine.LNX.4.64.0711271514180.5349@iabervon.org>
In-Reply-To: <FA6B0BE9-FC43-4298-932F-627826ADE18F@midwinter.com>

On Tue, 27 Nov 2007, Steven Grimm wrote:

Show 10 quoted lines

> On Nov 27, 2007, at 7:13 AM, Jakub Narebski wrote:
> >I thought that modern git refuses to push into checked out branch
> >(in HEAD) in non-bare repositories.
> 
> It doesn't -- otherwise the "update the working copy when a push to the
> current branch comes in" update hook scripts that some of us use wouldn't
> work, and they do work at the moment. (Before anyone warns me of the dangers
> of that: the hook only runs in a shared repo that no human is allowed to
> modify, so the working copy is always a clean version of HEAD and thus is safe
> to update.)

It could require the working copy to be a clean version of HEAD, and, in this case, update it. Then it would always be kept consistent one way or the other, without consistency depending on the use of a hook. (This is like how a pre-refs/remotes/ pull into a checked-out origin was handled, and it seemed reliable there)

	-Daniel
*This .sig left intentionally blank*