From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:57913) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Qoz2M-0000Be-42 for qemu-devel@nongnu.org; Thu, 04 Aug 2011 10:36:38 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Qoz2K-0000QP-V8 for qemu-devel@nongnu.org; Thu, 04 Aug 2011 10:36:38 -0400 Received: from mail-ww0-f53.google.com ([74.125.82.53]:40032) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Qoz2K-0000Q7-Nv for qemu-devel@nongnu.org; Thu, 04 Aug 2011 10:36:36 -0400 Received: by wwf26 with SMTP id 26so1782662wwf.10 for ; Thu, 04 Aug 2011 07:36:35 -0700 (PDT) Sender: Paolo Bonzini Message-ID: <4E3AAE6B.9030701@redhat.com> Date: Thu, 04 Aug 2011 16:36:27 +0200 From: Paolo Bonzini MIME-Version: 1.0 References: <1312326516-10117-1-git-send-email-aliguori@us.ibm.com> <4E389258.6040007@us.ibm.com> <4E398A44.8060507@codemonkey.ws> <4E3A97B5.6000002@redhat.com> In-Reply-To: <4E3A97B5.6000002@redhat.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH] migration: remove subsections in fdc and rtl8139 and bump versions List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Kevin Wolf Cc: Anthony Liguori , qemu-devel@nongnu.org, quintela@redhat.com On 08/04/2011 02:59 PM, Kevin Wolf wrote: >> > >> > I have. If we are changing a protocol in an incompatible version, we >> > can remove a lot of warts that current descriptions have. Not that >> > Paolo protocol is bad, but if we are going to do some change, adding >> > things like size, removing previous warts, etc is the way to go. > So how about stating clearly that migrating between 0.x and 1.x won't > work and using the next few months to develop a sane migration protocol? I think that's too early to state, perhaps we can keep backwards compatibility---who knows. And in case we totally screw up in 1.x, we still have my tweaked format as a backup. > We'll still have to do something about 0.15 and it's not very nice to > break migration twice, but seems there is no way around it. There is! Migration didn't break _that_ much with subsections, exactly because the buggy case occurs when subsections are sent and subsections are rare. If we always send them, as we did in practice with the floppy, we break stuff. That's why for 0.15 reverting the floppy fix is the simplest course of action. Paolo