From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Tue, 16 Nov 2004 14:54:38 +0000 (GMT) From: "Artem B. Bityuckiy" To: David Woodhouse In-Reply-To: <1100615381.8191.6957.camel@hades.cambridge.redhat.com> Message-ID: References: <1100613327.8191.6954.camel@hades.cambridge.redhat.com> <1100615381.8191.6957.camel@hades.cambridge.redhat.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: Artem Bityuckiy Cc: MTD List Subject: Re: your mail List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Tue, 16 Nov 2004, David Woodhouse wrote: > On Tue, 2004-11-16 at 14:21 +0000, Artem B. Bityuckiy wrote: > > I dreamed about checkpoints which we may write only once for constant > > files (for example, glibc libraries) .... > > > > So, this problem is solvable I believe, but no I'm not quite sure are > > checkpoints will be so useful as I thought ... > > Hm. I was imagining that we'd read in the information from the > checkpoint, and then pull in anything _new_ -- which would include any > nodes which weren't included in the checkpoint. Yes. But imagine some file which is constant and is never changed. We create checkpoint for it and this checkpoint covers all the nodes. So, we will need to rewrite new checkpoint even for this file when its nodes are garbage collected... -- Best Regards, Artem B. Bityuckiy, St.-Petersburg, Russia.