Re: [bfcpbis] Alexey Melnikov's Discuss on draft-ietf-bfcpbis-sdp-ws-uri-08: (with DISCUSS and COMMENT)

["Ram Mohan R (rmohanr)" <[e-mail]>]

Hi Alexey/all,

Please find the diffs with the changes to use single attribute. This diff also has other comments from Ben, Kathleen incorporated.

Regards,
Ram

-----Original Message-----
From: "Ram Mohan R (rmohanr)" <[e-mail]>
Date: Thursday, 19 January 2017 at 9:30 AM
To: Alissa Cooper <[e-mail]>
Cc: Alexey Melnikov <[e-mail]>, The IESG <[e-mail]>, "[e-mail]" <[e-mail]>, "[e-mail]" <[e-mail]>, "Charles Eckel (eckelcu)" <[e-mail]>, "[e-mail]" <[e-mail]>
Subject: Re: [bfcpbis] Alexey Melnikov's Discuss on draft-ietf-bfcpbis-sdp-ws-uri-08: (with DISCUSS and COMMENT)

    Hi Alissa,
    
    I don’t have a strong opinion on this. Either way (having a single attribute or two different SDP attributes) would work.
    
    Implementations can still retrieve the value (URI) of that attribute and based on the proto line (TCP/WS/BFCP or TCP/WSS/BFCP) match the URI against ws or wss schema defined in RFC6455.  I would expect such a validation to happen across layers.
    The SDP stack parser to look at the proto line transport value (TCP/WS/BFCP or TCP/WSS/BFCP) and check for presence of a=websocket-uri attribute. The value of this attribute (ws or wss schema) can be passed *as is* to the application (in this case BFCP client application) which will validate it against the schema in RFC 6455.
    
    I am OK to incorporate this comment to have a single SDP attribute. 
    
    Regards,
    Ram
    
    -----Original Message-----
    From: Alissa Cooper <[e-mail]>
    Date: Thursday, 19 January 2017 at 1:58 AM
    To: "Ram Mohan R (rmohanr)" <[e-mail]>
    Cc: Alexey Melnikov <[e-mail]>, The IESG <[e-mail]>, "[e-mail]" <[e-mail]>, "[e-mail]" <[e-mail]>, "Charles Eckel (eckelcu)" <[e-mail]>, "[e-mail]" <[e-mail]>
    Subject: Re: [bfcpbis] Alexey Melnikov's Discuss on draft-ietf-bfcpbis-sdp-ws-uri-08: (with DISCUSS and COMMENT)
    Resent-From: <[e-mail]>
    Resent-To: <[e-mail]>, <[e-mail]>
    Resent-Date: Thursday, 19 January 2017 at 1:58 AM
    
        Hi Ram,
        
        I had some further conversation with Alexey on the side and wanted to come back to a question he posed: is there a reason why a=ws-uri and a=wss-uri both need to be specified? Why can’t a single label be specified, say a=websocket-uri, and then have normative language requiring the scheme in the URI itself to match what’s in the associated m-line (ws:// for TCP/WS/BFCP and wss:// for TCP/WSS/BFCP)? Won’t it be more efficient for implementations to grab the URI itself anyway and check that the scheme matches the m-line?
        
        Thanks,
        Alissa
        
        > On Jan 17, 2017, at 9:40 AM, Ram Mohan R (rmohanr) <[e-mail]> wrote:
        > 
        > Alissa,
        > 
        > I am fine with the proposed text.
        > 
        > Thanks,
        > Ram
        > 
        > -----Original Message-----
        > From: Alissa Cooper <[e-mail]>
        > Date: Tuesday, 17 January 2017 at 8:07 PM
        > To: Alexey Melnikov <[e-mail]>
        > Cc: The IESG <[e-mail]>, "[e-mail]" <[e-mail]>, "[e-mail]" <[e-mail]>, "Charles Eckel (eckelcu)" <[e-mail]>, "[e-mail]" <[e-mail]>, "Ram Mohan R (rmohanr)" <[e-mail]>
        > Subject: Re: [bfcpbis] Alexey Melnikov's Discuss on draft-ietf-bfcpbis-sdp-ws-uri-08: (with DISCUSS and COMMENT)
        > 
        >    Alexey, would the following changes clarify things?
        > 
        >    Section 4.1
        >    OLD
        >    the server side, which could be either
        >       the offerer or answerer, MUST add an "a=ws-uri" or "a=wss-uri"
        >       attribute in the media section
        >    NEW
        >    the server side, which could be either
        >       the offerer or answerer, MUST add an "a=ws-uri" or "a=wss-uri”
        >       attribute (but not both) in the media section
        > 
        >    Section 4.3
        >    OLD
        >    If the answers assigns SDP "setup" attribute with "passive", then it
        >       MUST have a URI in either "a=ws-uri" or "a=wss-uri" attribute
        >       depending on whether the application uses WebSocket or
        >       secureWebSocket.
        >    NEW
        >    If the answers assigns SDP "setup" attribute with "passive", then it
        >       MUST have a URI in either "a=ws-uri" or "a=wss-uri" attribute
        >       (but not both) in the media section, depending on whether the application uses WebSocket or
        >       secureWebSocket.
        > 
        >    Ram, would you be okay with those clarifications?
        > 
        >    Alissa
        > 
        > 
        >> On Jan 16, 2017, at 10:09 AM, Ram Mohan R (rmohanr) <[e-mail]> wrote:
        >> 
        >> Hi Alexey,
        >> 
        >> -----Original Message-----
        >> From: Alexey Melnikov <[e-mail]>
        >> Date: Monday, 16 January 2017 at 8:25 PM
        >> To: "Ram Mohan R (rmohanr)" <[e-mail]>, The IESG <[e-mail]>
        >> Cc: "[e-mail]" <[e-mail]>, "[e-mail]" <[e-mail]>, "Charles Eckel (eckelcu)" <[e-mail]>, "[e-mail]" <[e-mail]>
        >> Subject: Re: [bfcpbis] Alexey Melnikov's Discuss on draft-ietf-bfcpbis-sdp-ws-uri-08: (with DISCUSS and COMMENT)
        >> 
        >>   Hi Ram,
        >> 
        >>   On Mon, Jan 16, 2017, at 02:49 PM, Ram Mohan R (rmohanr) wrote:
        >>> Hi Alexey,
        >>> 
        >>> Please see inline <Ram>
        >>> 
        >>> -----Original Message-----
        >>> From: Alexey Melnikov <[e-mail]>
        >>> Date: Monday, 16 January 2017 at 7:57 PM
        >>> To: "Ram Mohan R (rmohanr)" <[e-mail]>, The IESG <[e-mail]>
        >>> Cc: "[e-mail]"
        >>> <[e-mail]>, "[e-mail]"
        >>> <[e-mail]>, "Charles Eckel (eckelcu)" <[e-mail]>,
        >>> "[e-mail]" <[e-mail]>
        >>> Subject: Re: [bfcpbis] Alexey Melnikov's Discuss on
        >>> draft-ietf-bfcpbis-sdp-ws-uri-08: (with DISCUSS and COMMENT)
        >>> 
        >>>   Hi Ram,
        >>> 
        >>>   On Mon, Jan 16, 2017, at 06:09 AM, Ram Mohan R (rmohanr) wrote:
        >>>> Hi Alexey,
        >>>> 
        >>>> Thanks for your feedback. Please see inline <Ram>
        >>>> 
        >>>> -----Original Message-----
        >>>> From: bfcpbis <[e-mail]> on behalf of Alexey Melnikov
        >>>> <[e-mail]>
        >>>> Date: Saturday, 14 January 2017 at 3:49 AM
        >>>> To: The IESG <[e-mail]>
        >>>> Cc: "[e-mail]"
        >>>> <[e-mail]>, "[e-mail]"
        >>>> <[e-mail]>, "Charles Eckel (eckelcu)" <[e-mail]>,
        >>>> "[e-mail]" <[e-mail]>
        >>>> Subject: [bfcpbis] Alexey Melnikov's Discuss on
        >>>> draft-ietf-bfcpbis-sdp-ws-uri-08: (with DISCUSS and COMMENT)
        >>>> 
        >>>>   Alexey Melnikov has entered the following ballot position for
        >>>>   draft-ietf-bfcpbis-sdp-ws-uri-08: Discuss
        >>>> 
        >>> 
        >>>>   ----------------------------------------------------------------------
        >>>>   DISCUSS:
        >>>>   ----------------------------------------------------------------------
        >>>> 
        >>>>   This is generally a well written document, but I have a small list of
        >>>>   issues that I would like to discuss before recommending its approval:
        >>>> 
        >>>>   1) Are a=ws-uri and a=wss-uri mutually exclusive? (Section 4.3 is a
        >>>>   good
        >>>>   place to mention what to do if both are specified).
        >>>> 
        >>>> <Ram> Yes kind of. In a given media line (m= line) we will either have
        >>>> a=ws-uri or a=wss-uri. That said a response from a BFCP server using
        >>>> webSocket as a transport can
        >>>> have two media lines one with ws and other with wss. Something like:
        >>>> 
        >>>> Answer (server):
        >>>>  m=application 50000 TCP/WSS/BFCP *
        >>>>  a=setup:passive
        >>>>  a=connection:new
        >>>>  a=wss-uri:wss://bfcp-ws.example.com?token=3170449312
        >>>>  m=application 50000 TCP/WS/BFCP *
        >>>>  a=setup:passive
        >>>>  a=connection:new
        >>>>  a=ws-uri:ws://bfcp-ws.example.com?token=3170449312
        >>> 
        >>>   [Alexey]: This is exactly my problem, you should specify how a
        >>>   recipient
        >>>   should handle your example above. If you only have 1 attribute, the
        >>>   problem goes away entirely.
        >>> 
        >>> <Ram> in the above case, they are still two different m= lines. The
        >>> receiver (after offer/answer is done) of this SDP connects to the URI
        >>> mentioned in that m= line. 
        >>> Note I just re-used the same URL/port in the example above. If there are
        >>> multiple m=application media lines (they will be for different
        >>> applications) being negotiated in SDP, one application may use secure WS
        >>> and other application may use non-secure WS. The client after it receives
        >>> the answer SDP will just setup the connection to the URI specified in the
        >>> attribute.  
        >> 
        >>   Sorry, I was thinking about different example:
        >> 
        >>   m=application 50000 TCP/WSS/BFCP *
        >>      a=setup:passive
        >>      a=connection:new
        >>      a=wss-uri:wss://bfcp-ws.example.com?token=3170449312
        >>      a=ws-uri:ws://bfcp-ws.example.com?token=3170449312
        >> 
        >>   You have 2 conflicting URIs for the same m= line. Is this possible?
        >> 
        >> <Ram> No this is not possible. If the transport is TCP/WSS/BFCP we should have only a=wss-uri. If transport is TCP/WS/BFCP we should have only a=ws-uri.
        >> 
        >> Regards,
        >> Ram
        >> 
        >>> I don’t see any problem in here.  
        >>> 
        >>> Regards,
        >>> Ram
        >>> 
        >>>>   Why not a single attribute, considering that both ws: and wss: URIs
        >>>>   are
        >>>>   possible?
        >>>> 
        >>>> <Ram> I would still prefer two attributes. We just followed the
        >>>> convention / approach that was used in [RFC6455].  
        >>> 
        >>> 
        >> 
        >> 
        > 
        > 
        >