HE> This is caused by a non-resolved PN variable. PN is taken as "PN" HE> and not resolved to the string as in the #define. Yes, I just looked at the CPP output and I see that it's not correct. I have tried a bit, but I can't get the existing MIStub macros to behave like I want, since they assume that their "provider name" parameter is not a CPP macro :( ... HE> On the other hand, I sent out a patch to libcmpiutil some minutes HE> ago, which adds a STD_InstanceMIStub macro to libcmpiutil.h . This HE> macro avoids now the warning in the original stub macro, which HE> removes the responsibility from each provider to define it by HE> itself. The patches below show the difference (first one is yours HE> with PN approach, second one is removing the PN approach and using HE> STD_InstanceMIStub macro). Yeah, my goal with this was not to simply merge the name in the prototype and the macro, but rather to unify the name used for provider with multiple MIStub calls. If someone updates these in the future, they could update one and not the other and break things. However, I don't see an easy way around this. HE> Now that you touched each MI stub, this might also be a good point HE> in time to resolve this (now very old ;) ... sorry for my delay) HE> problem. Yeah, I can apply and re-roll this patch with just the required libvirt initialization and changes to use your updated macro. -- Dan Smith IBM Linux Technology Center Open Hypervisor Team email: danms@us.ibm.com