- From: Willy Tarreau <w@1wt.eu>
- Date: Tue, 21 Oct 2014 12:28:00 +0200
- To: Poul-Henning Kamp <phk@phk.freebsd.dk>
- Cc: Julian Reschke <julian.reschke@gmx.de>, Roberto Peon <grmocg@gmail.com>, Mark Nottingham <mnot@mnot.net>, RUELLAN Herve <Herve.Ruellan@crf.canon.fr>, Amos Jeffries <squid3@treenet.co.nz>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
On Tue, Oct 21, 2014 at 10:23:19AM +0000, Poul-Henning Kamp wrote: > -------- > In message <20141021102016.GG30397@1wt.eu>, Willy Tarreau writes: > > >We can have an encoding starting with something > >impossible in text mode (eg: LF byte) to mark [...] > > I would simply put two "Date:" entries in the static table, one > for text-encoded and one for number-encoded. That would not be compatible with dynamic indexing. I suspect that a number of applications emit many very similar dates in responses (when pages are dynamically generated), resulting in exact same values between the application and the frontend (lb/cache/etc). Given that you have much more experience than me looking at Date headers, don't you think that can be a loss ? Also it would require specific handling from the encoder to know that they must not index this specific header. We already have the set-cookie header which is boring, better try not to have another one. Just my few cents willy
Received on Tuesday, 21 October 2014 10:28:37 UTC