On 07/02/2010 07:45 AM, Chris Lalancette wrote:
Indeed, in the general case it *has* to be pushed to the individual driver entry points, because we don't know what flags individual drivers support. The best we could do is to do the check twice; at the libvirt entry point, we could check that it is one of the valid flags, and then at the driver entry point, check that it is one of the flags this driver supports. The benefit to this scheme is that we can catch gratuitious errors early without doing an RPC call; the downside is that we are repeating the virCheckFlags() call twice.
In any case, I think that would be a separate patch, so I would argue to leave this as-is for now (minus my grammatical mistake :).
Good argument - no change for now. And I _did_ see the virCheckFlags later on in 5/8 (it's just that I wrote my review immediately after reading 4/8, rather than reading the series). -- Eric Blake eblake@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org