On Thu, Jan 24, 2013 at 3:01 PM, Nico Williams <nico@cryptonector.com>wrote: > I mean, we have spent a great deal of effort over the last couple of > decades to minimize state in general, in TCP TCBs in particular, and > so on. We've done that for a reason. > > Also, there had better be a bound to stateful compression state size, > and therefore, deterministic synchronization or a synchronization > protocol (which would... add latency). > There is a bound, and the receiver (likely the server or proxy) gets to decide what it is. > > It's not at all obvious to me that stateful compression is a good > idea. It is clear that only stateful compression can do something > about values that get repeated a lot, like cookies and URL prefixes, > but I'm not sure that's worth the trouble. > Is saving seconds on page-load times worth the trouble? I ask because that is what we've seen in experiments... -=R > > Nico > -- >
Received on Thursday, 24 January 2013 23:05:15 UTC