Hi Juergen,
On 08/08/2022 12:31, Juergen Gross wrote:
And even with using an ID you'd have the same problem
again, but without having the possibility to add variant specific quota
Fair enough.
(remember that there already has been a statement that doing a live
update
from C to OCAML or vice versa would probably break users due to some
deviations in behavior)
I can't find such statement in public documentation. Do you have a link?
No, not really. IIRC this was a remark of Andrew back when we introduced
LU of Xenstore. It might even have been at the time the discussion was
only on the security ML.
Note that "users" doesn't need to imply guests, it might be related to
dom0 side users, like e.g. XAPI.
I understand that "users" doesn't imply guests. Hovewer, it is still not
quite clear to me what sort of behavior the application could rely on
here. Is it related to the transaction?
Anyway, I think it would be good to document such behaviors because they
don't seem to be obvious.
Cheers,
--
Julien Grall
