W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2014

Re: h2 use of Upgrade

From: Roy T. Fielding <fielding@gbiv.com>
Date: Tue, 2 Sep 2014 16:02:12 -0700
Cc: HTTP Working Group <ietf-http-wg@w3.org>
Message-Id: <C50F856F-D672-452E-9854-B776DFB35DA8@gbiv.com>
To: Martin Thomson <martin.thomson@gmail.com>
On Sep 2, 2014, at 3:18 PM, Martin Thomson wrote:

> On 31 August 2014 20:16, Roy T. Fielding <fielding@gbiv.com> wrote:
>> Note that the SETTINGS frame is a sufficient response if the request
>> was for OPTIONS.
> Are you suggesting that we wouldn't provide an HTTP-level response
> (other than the 101) to the OPTIONS request?

I am saying that the 101 (in HTTP/1.1 bits) is followed immediately by
a SETTINGS frame (in HTTP/2 bits), and that SETTINGS frame is a response
if the original request was for OPTIONS * (but not for something like
OPTIONS /).  OPTIONS * was designed for that purpose after the first
discussions of HTTPng.  SETTINGS is exactly what OPTIONS * is meant to
retrieve.  So, all we have to say is that an h2 response to an Upgrade
for an "OPTIONS * HTTP/1.1" request is the SETTINGS frame.

But that's just one alternative.  We could also simply tunnel the
HTTP/1.1 response back as a Headers+END_STREAM, which might be needed
for some unknown extensions if they already abuse OPTIONS *.

Received on Tuesday, 2 September 2014 23:02:35 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 30 March 2016 09:57:10 UTC