On 22-01-2015 14:45, Aline Manera wrote:
> Now I see that we cannot create that specific test scenario in
our
> mockmodel, given its current implementation. Our MockSnapshot is too
> simple, it isn't aware of the original VM's name (and many more
> things), so unless we change our class MockSnapshot - which I don't
> think it's worth it now - we cannot reproduce that issue. So we can
> keep it as it is now, as you proposed.
>
What are the consequences on it when running Kimchi on test mode?
Under the mockmodel, we're not able to reproduce the same scenario the
way it happens in the real model (i.e. on libvirt with QEMU), thus the
tests don't test the real behaviour. The class MockSnapshot doesn't keep
track of the original VM's name, so we cannot reproduce a situation
where reverting to a snapshot changes the VM name, which is what this
patch fixes. So the consequence is that the failing scenario cannot be
tested under our test mode, only if we reproduce that situation manuall
on a real model. Or unless we update the class MockSnapshot to be as
close as a real snapshot.