On Wed, Dec 23, 2015 at 07:02:01PM +0100, Paolo Bonzini wrote: > > > On 22/12/2015 18:24, Daniel P. Berrange wrote: > > I was a little reluctant to move this 'object_create' method into the > > qom/ code though, since I hate the idea of the legacy 'QemuOpts' data > > anywhere near those nice new APIs. I guess I could perhaps just keep the > > qemu_opts_to_qdict() call in the caller. > > I disagree with calling QemuOpts legacy, but I agree that this shouldn't > be in qom/object_interfaces.c. However, it could be moved to a separate > file in qom/.
Ok, yeah, I shouldn't have called it legacy really. This was my shorthand for saying that internal infrastructure APIs should strive to be qapi/qom based, and QemuOpts should only be used at the very top most level (ie cli ARGV) and converted to qapi/qom as soon as possible. I think that my v2 series achieves that goal fairly well. Regards, 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 :|
