From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:49744) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1clxGA-00024q-Dg for qemu-devel@nongnu.org; Thu, 09 Mar 2017 07:33:35 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1clxG6-0005so-Ck for qemu-devel@nongnu.org; Thu, 09 Mar 2017 07:33:34 -0500 Received: from mx1.redhat.com ([209.132.183.28]:57388) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1clxG6-0005sf-4s for qemu-devel@nongnu.org; Thu, 09 Mar 2017 07:33:30 -0500 From: Markus Armbruster References: <3d1c16a1-ec05-0367-e569-64a63b34f2e3@redhat.com> <940ff281-82cd-18cf-160e-c5234f65db18@redhat.com> <20170308112441.GI7470@redhat.com> Date: Thu, 09 Mar 2017 13:33:26 +0100 In-Reply-To: <20170308112441.GI7470@redhat.com> (Daniel P. Berrange's message of "Wed, 8 Mar 2017 11:24:41 +0000") Message-ID: <87bmtavdd5.fsf@dusky.pond.sub.org> MIME-Version: 1.0 Content-Type: text/plain Subject: Re: [Qemu-devel] What's the next QEMU version after 2.9 ? (or: when is a good point in time to get rid of old interfaces) List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Daniel P. Berrange" Cc: Thomas Huth , Peter Maydell , Jason Wang , QEMU Developers , Stefan Hajnoczi , Gerd Hoffmann "Daniel P. Berrange" writes: > On Wed, Mar 08, 2017 at 12:22:24PM +0100, Thomas Huth wrote: >> On 08.03.2017 11:03, Peter Maydell wrote: >> > On 8 March 2017 at 09:26, Thomas Huth wrote: >> >> But anyway, the more important thing that keeps me concerned is: Someone >> >> once told me that we should get rid of old parameters and interfaces >> >> (like HMP commands) primarily only when we're changing to a new major >> >> version number. As you all know, QEMU has a lot of legacy options, which >> >> are likely rather confusing than helpful for the new users nowadays, >> >> e.g. things like the "-net channel" option (which is fortunately even >> >> hardly documented), but maybe also even the whole vlan/hub concept in >> >> the net code, or legacy parameters like "-usbdevice". If we switch to >> >> version 3.0, could we agree to remove at least some of them? >> > >> > I think if we are going to deprecate and remove options we need >> > a clear transition plan for doing so, which means at least one >> > release where options are "still works, but warn that they >> > are going away with pointer to documentation or similar info >> > about their replacement syntax", before actually dropping them. >> >> Yes, that's certainly a good idea. But as Daniel suggested in his mail, >> I think we should also have the rule that the option should be marked as >> deprecated in multiple releases first - so that the users have a chance >> to speak up before something gets really removed (otherwise the option >> could be removed right on the first day after the initial release with >> the deprecation message, so there is no time for the user to notice this >> and complain). Not sure whether we need three releases, as Daniel >> suggested, though, but if that's common sense, that's fine for me, too. > > FYI, I didn't put any thought into my 3 releases / 12 months numbers. I > just arbitrarily picked them out of the hat, so don't consider it my > endorsement of that particular length of time :-) I think 2 is the minimum > number of releases we should deprecate for, beyond that, I'm open minded I don't think a hard rule for the grace period makes sense. It really depends. A guideline like "normally 12 months" is of course fine.