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

Re: review: http2-10

From: Martin Thomson <martin.thomson@gmail.com>
Date: Thu, 3 Apr 2014 13:21:12 -0700
Message-ID: <CABkgnnVP3sBSvwu3ZMghqmVEHod1hFqyF9Kr0UgHJ6NSEsBe0Q@mail.gmail.com>
To: Michael Köller <michael.koeller@greenbytes.de>
Cc: HTTP Working Group <ietf-http-wg@w3.org>, Julian Reschke <julian.reschke@greenbytes.de>
Thanks for calling out the shortcoming with respect to compression
statefulness.  I hope that this
(https://github.com/http2/http2-spec/commit/9939558e7c12bba8dc383ec294b059c98b8f8738)
helps clarify.

On 3 April 2014 12:36, Michael Köller <michael.koeller@greenbytes.de> wrote:
> Is there a reason that this frame type does not carry an optional debug data
> field, like GOAWAY does? I can image same situations and use cases on both.
> Speaking of debug data: Maybe the spec should advise about a reasonable
> default data format? Like a default encoding for text based data?

We discussed this possibility and decided not to add the debugging
data as a result of a combination of factors:

* Anything that requires debugging is probably worth elevating to
GOAWAY.  Any time RST_STREAM is used, the sender can send GOAWAY
instead.

 (Note that the working group has explicitly decided to avoid being
Postel-principle-compliant with this protocol: being lenient on errors
leads to people taking liberties, which leads to a giant mess.)

* There were DoS-related concerned with having unbounded data in
RST_STREAM, whereas the opportunity for exploitation on GOAWAY is
less.

* We only barely accepted the proposal in the first place.
Received on Thursday, 3 April 2014 20:21:40 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:14:29 UTC