Re: [rtcweb] Comment on draft-ietf-rtcweb-overview-02 (Browser RTC trapezoid)

["Ravindran Parthasarathi" <[e-mail]>]

Harald,

 

Please read inline.

 

Thanks

Partha

 

From: Harald Alvestrand [mailto:[e-mail]] 
Sent: Wednesday, October 19, 2011 10:48 AM
To: Ravindran Parthasarathi
Cc: [e-mail]
Subject: Re: Comment on draft-ietf-rtcweb-overview-02 (Browser RTC
trapezoid)

 

On 10/19/2011 01:12 AM, Ravindran Parthasarathi wrote: 

Harald,

 

In Fig 2 (Browser RTC Trapezoid), I'm getting the feel that JS/HTML
communicates with webserver directly without the involvement of browser.
Please clarify whether it is intended not to use browser  for JS.

Sorry, you completely lost me there.

The browser is the execution context for Javascript. The words "it is
intended not to use browser for JS" look like English to me, but I can't
find any interpretation that makes any sense.



<partha> Sorry, I messed up with the words here but you clarified it in
the below mail thread </partha>


If you don't have a clear view of the Javascript execution model, I
recommend spending a few hours with an introductory Javascript text and
playing with writing some Javascript in your own web pages. It will save
lots of confused emails to the list.

<partha> In fact I tried couple of W3Cschools Javascript before starting
the mail thread in RTCWeb. I'll learn more terminology in Javascript to
clearly present my ideas </partha>





 

IMO, the browser RTC trapezoid shall be 

 

                   +-----------+             +-----------+

                   |   RTCWeb  |  Federation |   RTCWeb  |

                   |           |  Signaling  |           |

                   |           |-------------|           |

                   |  Server   |   protocol  |  Server   |

                   |           |             |           |

                   +-----------+             +-----------+

                        /                           \

                       /                             \   RTCWeb

                      /                               \  Signaling

                     /                                 \

                    /  RTCWeb                           \

                   /   Signaling                         \

                  /                                       \

            +-----------+                           +-----------+

            |           |                           |           |

            |           |                           |           |

            |  Browser  | ------------------------- |  Browser  |

            |           |          Media path       |           |

            |           |                           |           |

            +-----------+                           +-----------+

            +-----------+                           +-----------+

            |JS/HTML/CSS|                           |JS/HTML/CSS|

            +-----------+                           +-----------+


Absolutely not.

1) The term "RTCWeb Signaling" has no clear definition, as this
discussion has proved again.

<partha> May I need to put clear definition by which we will come to
common understanding. </partha>


2) The conceptual difference between the media path (managed by the
browser without being mediated through Javascript) and the signalling
path (mediated through Javascript, transmitted through interfaces that
this WG has no intention of changing) is lost by the inversion of the
"browser stack".

 

<partha> Agreed and my intention is not manage media path through
Javascript. My figure has to change JS/HTML position to reflect it
correctly. </partha>



The detailed relationships between the browser components are better
described in Figure 1 of -overview- than in Figure 2 (the one you've
quoted). I've noted the need to mention that the JS uses browser
functions to access HTTP and WebSockets too, since it's apparently not
completely obvious to all.

<partha> My intention of the mail is above statement and it is not clear
from the figure 1 & 2.  </partha>






 




I explained the same diagram in Fig 1 of
draft-partha-rtcweb-signaling-00. RTCWeb signaling shall be proprietary
HTTP/websocket. I'm asking this change because I'm seeing folks confuses
API vs. on-wire-protocol. Also, Please note that JS + browser is the
single system with two different modules and it shall have protocol or
API to communicate (without passing any information in the wire). Could
you please let me know in case I'm missing something.

 

Thanks

Partha