Re: [hybi] I-D Action: draft-ietf-hybi-thewebsocketprotocol-11.txt

[Alexey Melnikov <[e-mail]>]

Takeshi Yoshino wrote:

> Hi Alexey,

Hi Takeshi,

> I have some questions about section 5.2.2. and 5.3.
>
> ----
>
> implied WSP rule -> implied LWS rule ?

Yes. It might have been called different things in different IETF RFCs, 
but I will double check and use the right term.

> ----
>
> Are 
> Sec-WebSocket-Protocol-Client, Sec-WebSocket-Version-Client, Sec-WebSocket-Protocol-Server, Sec-WebSocket-Version-Server 
> just names of ABNF rules?

Yes. The suffix "-Client" means that the syntax is used in HTTP requests 
(C->S), the suffix "-Server" means that the syntax is used in HTTP 
responses.

> i.e. Sec-WebSocket-Protocol-Client is the grammar for the right hand 
> side of the Sec-WebSocket-Protocol header in client handshake?

Yes.

> ----
>
> It seems header name and colon are missing from rules. e.g.
>
> Sec-WebSocket-Accept     = base64-value
>
> The left hand side was accept-value in -10 and the line seemed to be 
> defining the grammar of the right hand side of the 
> Sec-WebSocket-Accept data. So, just base64-value was fine. But now, 
> maybe you've tried to give grammar for full header line as RFC 2616 
> does and follow its style like
>        Accept         = "Accept" ":"
>                         #( media-range [ accept-params ] )
> so, this should be
>
> Sec-WebSocket-Accept     = "Sec-WebSocket-Accept" ":" base64-value

I suggest we do whatever RFC 2616 is doing. I will double check.

> ----
>
> This ABNF doesn't cover 1, 2, ... 99
>            version = "0" | ("1" DIGIT DIGIT) | ("2" DIGIT DIGIT)
>                        ; 0-255

Sigh, you are right, this needs to be fixed to match the comment.

> ----
>
> Maybe "1#" here is typo.
> Sec-WebSocket-Version-Server = 1#version

No typo here, this matches the text elsewhere in the document: the 
server advertises all versions it supports.