Paolo Bonzini <pbonz...@redhat.com> wrote: > On 06/15/2010 03:31 PM, Juan Quintela wrote: >> Signed-off-by: Juan Quintela<quint...@redhat.com> > > Sorry if this has been discussed to death before (if so I must have > missed it...). > > With subsections available, what about taking advantage of the new > protocol extension and add to the subsection info about the size of > the subsection?
Not trivial with current infrastructure :( > Also, with the size information, would it make sense to specify > optional subsections that the receiver could choose to ignore? We agreed that this was going to be forbidden. If sender send data -> it needs to be received. Sender can decide to not send a subsection if its data is not needed. > Mandatory subsections are something such that round-trip A->B->A fail > unless B understands the subsection, while optional subsections are > such that A can provide a default. IDE subsections would be optional, > for example. This is the whole point of the .needed() function. if neeeded(foo_subsection) returns false -> subsection is not needed. If it returns true -> it is needed, destination has to understand it. Later, Juan.
