于 2011年09月13日 18:07, Daniel Veillard 写道:
On Tue, Sep 13, 2011 at 05:18:40PM +0800, Daniel Veillard wrote:
On Tue, Sep 13, 2011 at 10:07:37AM +0100, Daniel P. Berrange wrote:
On Tue, Sep 13, 2011 at 01:08:00PM +0800, Daniel Veillard wrote:
Err, my example *was* still requiring that apps use 'import libvirt_qemu'
in order to get access to the QEMU specific APIs, and use that separate
objects to get their connection handle.
yes but from that point on if you use inheritance, the fact of using
the qemu tainted objects instead of the normal ones disapear completely.
The fact of using those temporary APIs get hidden in an import and a
new() somewhere. I would really prefer to see something explicit at the
place where it is used, something that people can't miss where reading
the code using it.
Okay, the two Dan disagree but none want to force the issue, can other
people voice in so that we can go with the majority :-) ?
Daniel
I have ever did a first version as what danpb suggested. But when talked
with DV on Friday morning, I was persuaed, so, I vote for defining the
function without class in libvirt_qemu.py, note that this will ease my
work. :-)
Regards,
Osier
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list