There are some proposals for why an intermediate peer might wish to use a compressed (opaque) via-list entry. Mostly to do with hiding some of the topology from other nodes.
To some extent, maybe the central question here is whether such behavior should be controlled by routing flags or overlay flags, and if a node wishes to hide topology, is it permissible for it to refuse to route messages inconsistent with this goal, thus forcing nodes trying to use other routing algorithms to fallback to more basic routing techniques. Roni, I think your suggestion of overlay parameters to restrict what routing options are available would accomplish the same goals as the routing flag (as far as enabling DRR-compliant forwarding), so I think if you think that's compatible with what you want to accomplish and are happy with it, we should go in that direction. Bruce On Wed, Feb 9, 2011 at 5:03 PM, Roni Even <[email protected]> wrote: > Hi, > I see what EKR means but I think that there is some inconsistency in section > 3.3. > The requirement says: > > "Low state: RELOAD's routing algorithms must not require > significant state to be stored on intermediate peers." > > Yet when talking about state keeping later in the section it says what EKR > points out > > "This option requires greater state to be stored on intermediate peers but > saves a small amount of bandwidth and reduces the need for modifying > the message en route. Selection of this mode of operation is a > choice for the individual peer;" > > I think that maybe this should be changed and that the selection of this > mode whould be allowed by overlay configuration and may be overridden by a > flag. I am not sure why an intermediary should decide on the via list > content. > > Roni Even > > >> -----Original Message----- >> From: Eric Rescorla [mailto:[email protected]] >> Sent: Wednesday, February 09, 2011 9:42 PM >> To: Roni Even >> Cc: 'Cullen Jennings'; 'P2PSIP WG' >> Subject: Re: [P2PSIP] WG decisions and issues on RELOAD base draft from >> meeting - DRR >> >> >> On Feb 9, 2011, at 11:31 AM, Roni Even wrote: >> >> > EKR, >> > I am not sure why the flag does not work for symmetric return, state >> > keeping by intermediaries is optional according to section 3.3 and is >> not >> > required for it to work. >> >> The intermediary's job is to keep the via list in a state where it is >> unwindable. It >> is a purely local matter how it does so. It's not reasonable to tell >> intermediaries >> who would otherwise keep state that they must instead modify the via >> list to >> include the full unwind path. >> >> -Ekr > > > _______________________________________________ > P2PSIP mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/p2psip > _______________________________________________ P2PSIP mailing list [email protected] https://www.ietf.org/mailman/listinfo/p2psip
