- From: Amos Jeffries <squid3@treenet.co.nz>
- Date: Wed, 28 Feb 2018 23:14:56 +1300
- To: ietf-http-wg@w3.org
On 28/02/18 17:30, Willy Tarreau wrote: > On Tue, Feb 27, 2018 at 09:06:32PM +0000, Poul-Henning Kamp wrote: >> -------- >> In message <MWHPR08MB2432213CA1E00C15FF346F11DAC00@MWHPR08MB2432.namprd08.prod.outlook.com>, Mike Bishop writes: >> >>> However, that means it's no longer possible to write the beginning >>> of the frame without having seen and processed the entire header >>> set. >> >> Speaking for the Varnish Cache project: We are fine with that. > > Same for haproxy, we always process all headers at once. > > And BTW, we have still not implemented support for CONTINUATION frames > because in the current model they would cause too many problems, > precisely because we process headers at once. No negative feedback for > this yet, only h2spec noticed, possibly indicating that they are not > that much used in field. > > Willy > Squid is the same in this regard. Always wait for the entire header block before any processing beyond parse happens. That is true regardless of HTTP version being used. Amos
Received on Wednesday, 28 February 2018 10:15:36 UTC