On 09/09/11 12:04, Kevin Wolf wrote: > Good to see agreement here. Do you think that Paolo's patches need to be > changed or can we do everything else on top?
A few things have come up on a third read, actually. I'll respond in due course to the appropriate patch. > We do have some timer stubs in qemu-tool.c since May. Of course, they > are just stubs and never really trigger the timer, but I assume that the > timeouts are really meant for qemu proper anyway, rather than qemu-img > or qemu-io. I wasn't aware of these - however, using them would make testing timeout/reconnect quite a bit harder. And surely there are use cases of qemu-img that would benefit from automatic reconnection? I can work around the former, and we don't actually use qemu-img for anything ourselves, so if you think this is the right way to go about it, I'm happy to rework the timeout/reconnect using QEMUTimer, on top of Paolo's patches. /Nick