- From: Martin Thomson <martin.thomson@gmail.com>
- Date: Wed, 16 Jan 2013 17:16:43 -0800
- To: Mark Nottingham <mnot@mnot.net>
- Cc: Nico Williams <nico@cryptonector.com>, Roberto Peon <grmocg@gmail.com>, James M Snell <jasnell@gmail.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
I'm less concerned about that sort of error than I am with getting epoch right in the first place. Though I note that you would want this per-stream and possibly per-direction to allow for proper, translation free multiplexing at intermediaries. That seems like more work than can be justified for a very modest saving over the already significant gain. On 16 January 2013 16:47, Mark Nottingham <mnot@mnot.net> wrote: > On 17/01/2013, at 11:25 AM, Nico Williams <nico@cryptonector.com> wrote: > >> On Wed, Jan 16, 2013 at 6:04 PM, Roberto Peon <grmocg@gmail.com> wrote: >>> If we're doing compression, we're associating state with connections anyway. >> >> As long as it's transparent and hop-by-hop, it works. That would work >> dates, huh. > > > The other factor is that if it's connection-specific, each hop will have to keep that state and properly translate it to the next one, introducing the possibility for errors. > > > -- > Mark Nottingham http://www.mnot.net/ > > > >
Received on Thursday, 17 January 2013 01:17:10 UTC