[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [Libvir] os/bootloader



On Sun, Dec 09, 2007 at 12:32:06AM +0000, Daniel P. Berrange wrote:

> >           <group>
> >             <ref name='os'/>
> >             <optional>
> >               <ref name='bootloader'/>
> >             </optional>
> >           </group>
> >           <group>
> >             <ref name='bootloader'/>
> >             <optional>
> >               <ref name='os'/>
> >             </optional>
> >           </group>
> > 
> > means I can specify either os, or bootloader, or both. The latter case
> > makes a lot of sense when using pygrub. However, src/xml.c doesn't
> > implement those semantics:
> > 
> > 1219     if (!bootloader) {
> > 1220         if ((node = virXPathNode("/domain/os[1]", ctxt)) != NULL) {
> > 1221             /* Analyze of the os description, based on HVM or PV.  */
> > 1222             str = virXPathString("string(/domain/os/type[1])", ctxt);
> > 
> > 
> > I think that we should just remove the "if (!bootloader) {" bit here -
> > agree?
> > 
> > I'm looking at the schema and I don't see (at least)
> > domain/devices/input, which appears to be used in src/xml.c too - is the
> > schema out of date in 0.3.3 ? The tests don't pass xmllint!
> 
> Sorry, yes its out of date wrt to this. If the schema doesn't match the 
> code, then the schema is wrong. The code is definitive (at least for
> the Xen / KVM drivers).

I presume this applies to the second complaint, not the first?

regards
john


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]