Re: Getting to Consensus: CONTINUATION-related issues

I prefer B, by knowing the uncompressed limit the application layer - where
the messages are built - adjustments can be made if it's going over the
limit.

If A is selected I'll most likely either throw away the advertised limit
and hope the remote likes what I send, or try to somehow derive a sensible
uncompressed limit from the compressed limit to pass up to the application
layer. It's unreasonable to expect the application layer to stay under a
limit that depends on the order of delivery.

I'm not really concerned about continuations per se.