On 04/16/2016 11:12 AM, Wouter Verhelst wrote: > On Fri, Apr 15, 2016 at 11:39:12AM +0100, Alex Bligh wrote: >> >> On 15 Apr 2016, at 08:09, Wouter Verhelst <w@...112...> wrote: >> >>>> If a server advertises a minimum >>>> +block size, the advertised export size MUST be an integer multiple of >>>> +that block size. >>> >>> I think this can be a SHOULD without problem? >> >> Actually I think this should be a MUST. If the server is imposing >> restrictions, they should be consistent. > > Which is why I said SHOULD, not MAY. > >> The client actually can't access the export beyond the integer multiple >> without violating the spec, so in my mind it's up to the server to round the >> size down. > > Sure. My point is that the client would have to have a <= test anyway, > in which case it doesn't matter if the size isn't a nice integer > multiple. It's not proper, and we should discourage people from doing it > wrong (which is what SHOULD does), but it doesn't cause many problems if > you did it wrong anyway. Okay, I think I'll word it as SHOULD, with the caveat that a server violating the SHOULD has made it impossible for the client to access the tail of the file. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
Attachment:
signature.asc
Description: OpenPGP digital signature