From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1Lte0R-0005t5-3D for qemu-devel@nongnu.org; Tue, 14 Apr 2009 04:28:35 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1Lte0M-0005qK-Ea for qemu-devel@nongnu.org; Tue, 14 Apr 2009 04:28:34 -0400 Received: from [199.232.76.173] (port=41727 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1Lte0M-0005q9-20 for qemu-devel@nongnu.org; Tue, 14 Apr 2009 04:28:30 -0400 Received: from mx2.redhat.com ([66.187.237.31]:48746) by monty-python.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1Lte0K-00049b-DM for qemu-devel@nongnu.org; Tue, 14 Apr 2009 04:28:29 -0400 Message-ID: <49E44920.8000600@redhat.com> Date: Tue, 14 Apr 2009 10:28:16 +0200 From: Gerd Hoffmann MIME-Version: 1.0 Subject: Re: [libvirt] Re: [Qemu-devel] [PATCH 1/6] Allow multiple monitor devices (v2) References: <1239215702-23818-1-git-send-email-aliguori@us.ibm.com> <49DDAF9F.7040400@redhat.com> <49DDF807.1050707@us.ibm.com> <49DDFAD5.7060808@redhat.com> <49DDFC5C.4080504@us.ibm.com> <49DE0042.9050103@redhat.com> <49DE0271.8090103@us.ibm.com> <49DE05F2.5060304@redhat.com> <49DE0673.3070501@us.ibm.com> <49DE081D.1030702@redhat.com> <49DE0CF2.3060307@us.ibm.com> <49DE1029.3030909@redhat.com> <49DE16DE.6030206@us.ibm.com> <49DE1AD2.2060600@redhat.com> <49DE1DB3.30402@us.ibm.com> <49DE229B.7050408@redhat.com> <49DE2580.2030507@us.ibm.com> <49DE29C5.6010601@redhat.com> <49DE3321.4090900@us.ibm.com> <49E0C47F.9070501@redhat.com> <49E124BB.1010604@redhat.com> <49E1A517.50105@redhat.com> In-Reply-To: <49E1A517.50105@redhat.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Reply-To: qemu-devel@nongnu.org List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: libvir-list@redhat.com, Anthony Liguori , Jan Kiszka , Avi Kivity , Hollis Blanchard Hi, > Personally, I think relying on asynchronous events to provide reliable > status is a bad idea. Management connections can and will get > disconnected, buffers will get filled and event notifications will be > dropped. I somehow like the idea from someone (Jamie?) somewhere in this thread to model the notification like unix signals. Just flag that something changed. Don't carry the actual info, just ping the management app, which in turn must use "info " to get the details. That solves the queuing issue. I also closes the command vs. notification races. Management apps don't need new parsers, the existing "info " parser(s) will do just fine. cheers, Gerd