Re: [Libvirt-cim] [PATCH] Some minor std_association fixes

KR> You mean, instead of returning success, we should set the status KR> using cu_statusf(). Returning an error makes sense from a client KR> point of view. If the client thinks an association is going to KR> return a specific instance, and nothing is returned, it won't know KR> the state of the system. If an error is returned, it's much more KR> obvious to the client what has happened. Right, if you return OK, but no instances, then it just looks like there is nothing on the other end of your association, which would normally be a transient state. However, asking for a result class that we don't support is not a transient error, it's a real one, and we should signal as such. Thanks for catching this and bringing it into the foreground for the requisite beating :) -- Dan Smith IBM Linux Technology Center Open Hypervisor Team email: danms(a)us.ibm.com