Re: Clarification on HTTP/1.1 Server Response to HTTP/2 Client Connection Preface from Martin Thomson on 2014-08-28 (ietf-http-wg@w3.org from July to September 2014)

From: Martin Thomson <martin.thomson@gmail.com>
Date: Thu, 28 Aug 2014 11:39:51 -0700
To: Lucas Pardue <Lucas.Pardue@bbc.co.uk>
Cc: Mike Bishop <Michael.Bishop@microsoft.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
Message-ID: <CABkgnnXb=Qrie=wMQUGvkA76P_TExC8grOzPNgzMAKVaP6g7Ag@mail.gmail.com>

On 28 August 2014 08:56, Lucas Pardue <Lucas.Pardue@bbc.co.uk> wrote:
> However, hypothetically it may contain special handling for the request line
> and could correctly issue a 505 via HTTP/1.1, with enhancement.

I'll point out that there's nothing correct about the preface we send,
so any behaviour that is immediately obvious as a failure, like
dropping the connection, or something like 505, is OK here.  The key
is to avoid misleading the client into thinking that its assumption of
HTTP/2 support was correct.  Generating a valid HTTP/2 GOAWAY, with
all the lead-in, is probably the best way to avoid this, assuming that
the client understands the code we choose.  (Mike, that's an argument
for resolving this issue now rather than later; I'll follow up on
that.)

Received on Thursday, 28 August 2014 18:40:19 UTC