Re: WebSocket over HTTP2 RFC6455 conformance

On Thu, Feb 20, 2014 at 5:59 AM, Martin Thomson <martin.thomson@gmail.com>wrote:

> On 19 February 2014 06:11, Yutaka Hirano <yhirano@google.com> wrote:
> > At hybi, the following people answered "yes & yes".
>
> When you ask about semantics, you need to be more precise.
>

Good point, Martin. Here're HyBi threads started recently and calling for
opinions for smaller primitive questions.

Is it important to know frame length at the start of frame?
http://www.ietf.org/mail-archive/web/hybi/current/msg10404.html

Straw poll: Do you miss per-frame extension?
http://www.ietf.org/mail-archive/web/hybi/current/msg10448.html

Straw poll: Do you miss interjectable WebSocket level control frame?
http://www.ietf.org/mail-archive/web/hybi/current/msg10450.html


> My interpretation of "Websocket semantics" is the set of things that
> you can do with thewebsocketprotocol, or maybe the W3C WebSocket API.
> If your intent is to tunnel the protocol in a completely lossless
> fashion, then you are asking for more than just a preservation of
> semantics, you also want to preserve syntax.
>

Yutaka didn't mean to include syntax (I believe you meant binary
serialization), but preserving frame boundary, etc.


> Note that we preserve syntax in HTTP/2 because there is a wealth of
> usage out there that relies on syntax, and even if we might think it
> unwise to do so, we are still unwilling to break those users.  It
> might be that the hybi group has people using those reserved bits such
> that preserving them is paramount, but websockets has far less history
> than HTTP.  I also didn't find the framing of the question or the
> responses to be particularly convincing.
>

We're not pushing HyBi's decision but want to ask httpbis people's opinions
as well. Sorry if you felt like that.

So, you're also not interested in preserving RSV1-3 as-is even as
per-message basis?