On 07/24/2012 09:58 AM, Daniel P. Berrange wrote: > On Tue, Jul 24, 2012 at 09:16:38AM +0200, Martin Kletzander wrote: >> On 07/23/2012 09:42 PM, Daniel P. Berrange wrote: >>> On Mon, Jul 23, 2012 at 02:25:23PM -0500, Doug Goldstein wrote: >>>> On Mon, Jul 23, 2012 at 1:51 PM, Martin Kletzander <mkletzan@xxxxxxxxxx> wrote: >>>>> The 'domdisplay' command didn't properly evaluate '--include-password' >>>>> option. >>>>> --- >>>>> tools/virsh.c | 35 +++++++++++++++++++++++------------ >>>>> 1 files changed, 23 insertions(+), 12 deletions(-) >>>>> >>>>> diff --git a/tools/virsh.c b/tools/virsh.c >>>>> index 5888d6c..e0765ba 100644 >>>>> --- a/tools/virsh.c >>>>> +++ b/tools/virsh.c >>>>> @@ -66,6 +66,7 @@ >>>>> #include "virtypedparam.h" >>>>> #include "intprops.h" >>>>> #include "conf/virdomainlist.h" >>>>> +#include "datatypes.h" >>>> >>>> Why exactly is this necessary? No new types are being used that aren't >>>> used throughout this whole file. >>> >>> Indeed, datatypes.c is for the libvirt private details of >>> the public objects. virsh is a public API consumer, so has >>> no justification for using this include. >>> >> >> I wasn't able to dereference virConnectPtr (conn->flags) without this >> include, so gcc ended with an error. > > virsh has no business touching conn->flags at all. That is private > libvirt data, not for public consumption. > That makes sense. I didn't know where exactly get the flags (dom->conn->flags maybe?), but it won't be in v2 anyway. Martin -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list