Re: [rtcweb] How to multiplex between peers

[Igor Faynberg <[e-mail]>]

Maybe, at some point.

At the moment though I am not convinced that this is the right answer.  
It always looks easier to re-design everything, but this approach more 
often than not ends up in a standard (and implementations) full of 
bugs.  Looking at what has transpired in Transport (now RAI) in the past 
15 years, one would easily find examples.

I think the decision on whether to ignore backward compatibility ought 
to be made when there is an agreement on core use cases and it is 
understood that none of them would be broken by lack of backward 
compatibility.

(It is not at all that I am against re-design on principle.  Quite the 
opposite!  But I believe that re-design is the research issue; here, we 
write a standard, and a standard needs to take into account a plethora 
of issues that the research phase may and should ignore.)

Igor

On 7/20/2011 7:15 PM, Matthew Kaufman wrote:
> ...
>
> Could we confine the claims about breakage to only ones that are NOT backwards compatibility issues?
>
> ...