Re: [bfcpbis] reminder to review draft-ietf-bfcpbis-bfcp-websocket-01

["Gonzalo Salgueiro (gsalguei)" <[e-mail]>]

Hi Christer - 

We have submitted a new version of the draft (-03) that addresses your remaining comments.  This includes separating the WS/WSS Connection URI definition and registration to a separate draft (See https://tools.ietf.org/html/draft-ram-bfcpbis-sdp-ws-uri-00).

At your convenience please have a look and see if your comments are satisfactorily addressed by both drafts.

Thanks!

Gonzalo


> On Oct 25, 2014, at 7:40 AM, Christer Holmberg <[e-mail]> wrote:
> 
> Hi,
> 
>>> Section 3:
>>> ------------
>>> 
>>> Q3_1: Why is this section needed? The draft is not a teach-yourself-websocket document, is it? :)
>>> 
>>>          In my opinion the draft should reference the WebSocket RFC for general information about WebSocket, and 
>>>        the draft then describes the BFCP specifics - ie what is described in section 4.
>>> 
>>>          Otherwise, if some details of WebSocket change in the future, you may have to update this RFC 
>>>          also - even if the change does not affect the BFCP usage of WebSocket.
>> 
>> A bit of explanatory text was requested in early reviews.  I myself find it useful to provide modest 
>> baseline information self-contained in the document.  This is also common practice in all the foo 
>> over WebSockets documents, including RFC 7118.  That said, if you feel strongly it should be removed, we can do so. 
> 
> If you want to keep it, I will withdraw my comment. It's just an editorial thing :)
> 
> ...
> 
>>> Q6_2: Is the new SDP ws-uri attribute BFCP specific? If not, should it be in a separate draft?
>> 
>> No, it isn't BFCP-specific and your point makes sense.  Having a general SDP attribute for SDP negotiated 
>> protocols over Websocket is meaningful and would facilitate future protocols over Websocket. Once I create 
>> a new draft I'll remove this bit of text and replace it with a reference to the new document.  Expect this for the -03 version.
> 
>>> Q6_3: I think it would be good to have a reference to a document which talks more about the URI matching requirement.
>>> 
>>>          Something like: "As defined in section YYYY of RFC XXXX, when using Secure Websockets the CNAME 
>>>          of the SSL certificate must match the WebSocket connection URI host..."
>> 
>> I agree with the comment, but not sure what the reference should be. Not sure what will be the correct reference. 
>> Do you think something like RFC 2818 Section 3 would work?
> 
> I don't know where the requirement comes from, but I assume it must be defined somewhere :)
> 
> Anyway, if you move the SDP ws-uri attribute definition into a separate draft I assume this text would also be in that draft.
> 
>>> Q6_4: As you define a new SDP attribute, you should have a "SDP Offer/Answer Proceudures" section.
>> 
>> Do you mean an Offer/Answer sections in line with RFC 3264 that has the following sub-sections with relevant procedures called out?
>> 
>> 1) Generating the Initial Offer
>> 2) Generating the Answer
>> 3) Offerer Processing of the Answer
>> 4) Modify the Session
>> 
>> Something like what we did in Section 4 of RFC 7345?
> 
> Correct.
> 
> Again, if you move the attribute definition to a separate draft, you only need to define the O/A procedures for the new m- line transport field values in this draft.
> 
>> If so, that will take me a bit of time and will need to wait until -03 as I have much to do before the submission deadline.
> 
> That is ok. I guess you could put the section names into -02, and add the content later.
> 
> Regards,
> 
> Christer