Re: [hybi] Compression spec decoupling

[Greg Wilkins <[e-mail]>]

I prefer a) to b),  but don't strongly object to b)

cheers



On 26 April 2012 19:11, Arman Djusupov <[e-mail]> wrote:
> I’m also fine with both options.
>
>
>
> With best regards,
>
> Arman
>
>
>
> From: Takeshi Yoshino [mailto:[e-mail]]
> Sent: Thursday, April 26, 2012 7:24 AM
> To: [e-mail]; John Tamplin; Arman Djusupov; Greg Wilkins
> Subject: Compression spec decoupling
>
>
>
> As the target data is close, I'd like to list options with concrete examples
> and call for opinions to settle this topic.
>
>
>
> See also this
> thread: http://www.ietf.org/mail-archive/web/hybi/current/msg09469.html
>
>
>
> We have two options.
>
> (I saw slight objection to add a new handshake header to pass
> algorithm parameter in f2f meeting)
>
>
>
> a) One per-frame compression extension for all compression algorithm
>
>
>
> draft-ietf-hybi-websocket-perframe-compression defines
>
> - an extension "perframe-compress" and how it frames data including usage of
> RSV1
>
> - an extension parameter "algorithm" and its format
>
> - an algorithm parameter "deflate" and how it transforms data
>
>
>
> RSV1 is given to perframe-compress extension. It might make incompatible
> extension checking easier.
>
>
>
> There something does "per-frame compression" but needs framing beyond what
> we define in this spec might come in the future. Then, we need to choose
> either of break this 1:1 assignment or forbid it to use RSV1.
>
>
>
> Example:
>
> Sec-WebSocket-Extensions: perframe-compress; algorithm="foo; foo-param1=123,
> bar; bar-param1=abc"; general-param1=xyz", ...
>
>
>
> Example (with mux):
>
> Sec-WebSocket-Extensions: perframe-compress; algorithm="... (algorithms for
> pre-mux) ...", mux, perframe-compress; algorithm="... (algorithms for
> post-mux) ..."
>
>
>
> b) One extension for each per-frame compression algorithm
>
>
>
> draft-ietf-hybi-websocket-perframe-compression defines
>
> - a method how to frame data with per-frame compression including usage of
> RSV1
>
> - an extension "deflate-frame" which uses the framing method and how to
> transforms data
>
>
>
> RSV1 is given to any extension that does per-frame compression. Endpoints
> need to sort out mutually exclusive extensions.
>
>
>
> Any future per-frame compression extension may or may not refer to the
> framing method defined in draft-ietf-hybi-websocket-perframe-compression.
>
>
>
> Example:
>
> Sec-WebSocket-Extensions: foo-compress; foo-param1=123, bar-compress;
> bar-param-1=abc, ...
>
>
>
> Example (with mux):
>
> Sec-WebSocket-Extensions: ... (extensions for pre-mux) ...", mux, ...
> (extensions for post-mux) ..."
>
>