Re: dangling trees

From: Linus Torvalds <torvalds@osdl.org>
Date: May 5, 2005, 14:41 UTC
Message-ID: <Pine.LNX.4.58.0505050738150.2328@ppc970.osdl.org>
In-Reply-To: <20050505141224.A8323@flint.arm.linux.org.uk>

On Thu, 5 May 2005, Russell King wrote:

Show 10 quoted lines

>
> In todays 2.6 git repo, I see the following:
> 
> $ fsck-cache
> expect dangling commits - potential heads - due to lack of head information
> dangling commit 897f5ab2cd733a77a2279268262919caa8154b9d
> dangling tree c39ae07f393806ccf406ef966e9a15afc43cc36a
> 
> The dangling commit is the current head - that's fine.  However,
> what's this dangling tree?

That's the 2.6.11 tree, which I added to have a previous real release to diff against.

> It appears to have come in from Linus' tree on kernel.org this morning.

Yes. You can make (my) fsck shut up with something like

	refs=.git/refs/*/*
	git-fsck-cache $refs

but that assumes that cogito has been updated to the newer fsck that understands to follow tags (not just parse them, but also add the "tag refers to object xxx" info to its internal tree of trust). I don't think that has happened yet.

In the meantime, just ignore it.

		Linus