On Wed, Oct 6, 2021 at 6:13 AM Paolo Bonzini <pbonz...@redhat.com> wrote:
> On 23/09/21 02:49, John Snow wrote: > > This is a wrapper around the async QMPClient that mimics the old, > > synchronous QEMUMonitorProtocol class. It is designed to be > > interchangeable with the old implementation. > > > > It does not, however, attempt to mimic Exception compatibility. > > > > Signed-off-by: John Snow <js...@redhat.com> > > Acked-by: Hanna Reitz <hre...@redhat.com> > > I don't like the name (of course). qemu-iotests shows that there is a > use for sync wrappers so, with similar reasoning to patch 16, there's no > need to scare people away. Why not just qemu.aqmp.sync? > > Paolo > > I had plans at one point to make a sync.py, but with an interface that matched async QMP itself more closely. I spent some time trying to research how to make a "magic" sync wrapper around async QMP, and hit a few trouble spots. I've still got the patch, but I felt some pressure to try and switch iotests over as fast as possible to get more trial-by-fire time this release cycle. I named them "sync.py" and "legacy.py" in my branch accordingly. Of course, I made a beeline straight for the iotests version, so now it looks odd. I may yet try to clean up the other version, possibly converting legacy.py to work in terms of sync.py, and then converting users in iotests so that I can drop legacy.py. (Mayyyybe. I am not heavily committed to any one particular approach here, other than being very motivated to get AQMP tested widely a good bit before rc0 to have a chance to smooth over any lingering problems that might exist.) Thanks for taking a look! I am more than happy to stage 1-9 myself, but I will wait for Hanna's approval on 10-14 in particular. --js
