- From: Mike Bishop <Michael.Bishop@microsoft.com>
- Date: Tue, 25 Jul 2017 16:37:08 +0000
- To: Willy Tarreau <w@1wt.eu>
- CC: Ilari Liusvaara <ilariliusvaara@welho.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
No problem, I know it's a little weird. The logic is that, because the HEADERS frame can't be processed individually, HEADERS+CONTINUATION get treated as a single very large HEADERS frame. -----Original Message----- From: Willy Tarreau [mailto:w@1wt.eu] Sent: Monday, July 24, 2017 10:15 PM To: Mike Bishop <Michael.Bishop@microsoft.com> Cc: Ilari Liusvaara <ilariliusvaara@welho.com>; ietf-http-wg@w3.org Subject: Re: Issue with the H2 state machine Hi Mike, On Mon, Jul 24, 2017 at 05:16:42PM +0000, Mike Bishop wrote: > No, because that would prohibit ridiculously large trailers (i.e. > using CONTINUATION), which might be stupid but are permitted. The > spec is saying that any HEADERS frame after the final status code has > to be trailers, and therefore that block is the last thing on the > stream. Alternatively stated, you can't send your trailers but then send more body. Thanks, now I understand my mistake. I did not realize that the ES bit had to be put on the HEADERS frame even if there are CONTINUATION frames but now I found where it's mentionned and all of this makes sense. It'll require some changes to my state machine, but that's exactly why I prefered to be sure to be able to deal with CONTINUATION frames from the start ;-) Cheers, Willy
Received on Tuesday, 25 July 2017 16:37:33 UTC