On Tue, 09 Mar 2010 08:48:43 -0600 Adam Litke <a...@us.ibm.com> wrote:
> On Tue, 2010-03-09 at 11:22 -0300, Luiz Capitulino wrote: > > On Tue, 09 Mar 2010 14:51:31 +0100 > > Juan Quintela <quint...@redhat.com> wrote: > > > > > Any recompilation/etc would break migration. I have tried to understand > > > what happened with monitor async commands, and my head exploded in > > > indirections. > > > > The Monitor needs lots of cleanups to make things more obvious. > > > > > Is there any written explanation of what are we trying to do here? > > > > Only the commit log 940cc30. > > > > Basically, an asynchronous handler has a completion function which is > > called when the handler completes. > > > > If we're in the user Monitor, it's suspended until the completion > > function is called. In QMP, the handler returns immediately and we > > _should_ be emitting an event when we have the answer. > > > > The current code doesn't do that, which seems to be a new issue. > > With current git, I cannot get QMP to recognize any commands. Unless > this is a known issue, I will look into it further to see what has > caused it. You have to issue the 'qmp_capabilites' command before issuing commands, take a look at the QMP/README and QMP/qmp-spec.txt files.
