> On Feb 26, 2026, at 08:02, Robert Raszuk <[email protected]> wrote: > > Hi Jeff, > > Let me present how I understood the meaning of this bit. Note that the term > "overload" is poorly chosen. > > Sender is sending Overload == 0 --> Receiver's action: For all matching > routes (to sent filter) in Ajd-RIB-OUT withdraw them to the peer. > > Sender is sending Overload == 1 --> Receiver's action: For all new routes > (to sent filter) not yet present in Ajd-RIB-OUT suppress from sending to the > peer.
If that's the intended behavior, I think that's clear. Essentially, when bit is 1, do not withdraw but "send nothing further". If AIjun agrees about the behavior, we can adjust the English in the draft. > > Yes in that context sequence matters, but this IMO would be really a corner > case and pretty ugly way to use sequence numbers. This is partially my point. > > Instead draft can mandate to never sent mutually overlapping filters with > different overload bits. I know that's your personal preference. And yet, an implementation that receives such a conflicting ORF must do something consistent with it. Ordering is a reasonable tie-breaking criteria. -- Jeff _______________________________________________ BESS mailing list -- [email protected] To unsubscribe send an email to [email protected]
