On Mon, Oct 08, 2012 at 04:48:00PM +0200, Jiri Denemark wrote:
On Mon, Oct 08, 2012 at 15:35:38 +0100, Daniel P. Berrange wrote:
> On Mon, Oct 08, 2012 at 04:30:03PM +0200, Jiri Denemark wrote:
> > QEMU changed the output of -cpu ? and CPU models can now be followed by
> > its description. Our parser just used both model name and its
> > description as a model name, which made any model with a description
> > unusable with libvirt.
>
> What version of QEMU changed this ?
Hmm, I think I see why you are asking... I just checked and the changset that
introduced this is v1.2.0-266-g6cdf885, with which we will use QMP probing.
Thus, we don't need this code for working with upstream qemu. (Not that this
would be the first case of committing a patch that makes libvirt work better
with downstream qemu with backported patches.)
My feeling is that we came to agreement with the QEMU developers that we
would exclusively useful QMP probing with any QEMU from 1.2.0 or later
and would not try to follow changes they make to the non-QMP interfaces
we previously relied on. So if there are any changes to -help, -cpu, -M
or -device output from 1.2.0 onwards, we really should not try to follow
them in libvirt master.
I could see this patch being acceptable for our newest stable release
stream branch though. Traditionally our requriement was that patches
for stable should go in master first, but I think this scenario is an
example where we should relax that rule and put the change in the stable
branch only.
Daniel
--
|:
http://berrange.com -o-
http://www.flickr.com/photos/dberrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|:
http://entangle-photo.org -o-
http://live.gnome.org/gtk-vnc :|