On Wed, 11 Apr 2012, Christopher Morrow wrote:
"if you don't ask for the 'bgpsec capability' then ... you get what you get today."
so, everything you do today, ought to just keep right on working, or that's the plan.
Capability negotiation does not mean everything keeps on working. It means the session between the BGP speakers keeps working, sure. However, that's _not_ everything.
The BGP UPDATE message is no longer context-complete (wrt to the well-known attributes at least). If a 3rd-party wishes to be able to validate the message as cortrect (in the case of missing attributes) or even decode it correctly (in the case where well-known attributes are incompatibly redefined in syntax, like AS_PATH has been), it has to have seen the opening negotiation - which may have happened days or weeks or more before - or it has to be manually configured or make intelligent guesses.
It should be quite possible to keep BGP as completely parse-able at the message granularity - it just needs a modicum of care. It's a shame that ever more proposals are coming along that are overloading message formats, dependent on context that may only be exchanged very infrequently...
regards, -- Paul Jakma p...@jakma.org @pjakma Key ID: 64A2FF6A Fortune: static from plastic slide rules _______________________________________________ sidr mailing list sidr@ietf.org https://www.ietf.org/mailman/listinfo/sidr
