Re: [rtcweb] Plan A, respun

The Iraqi Information minister did not correctly communicate "facts on the ground", whereas I do think Harald correctly characterizes how some existing implementations behave. 

Given the discussion on this topic I would suggest that it would be useful to include a recommendation for WebRTC implementations (in the RTP usage document?).

On May 16, 2013, at 11:27, "Harald Alvestrand" <harald@alvestrand.no> wrote:

> Adam, this is the first time I've been compared to the Iraqi Information Minister... another achievement unlocked!
> 
> There are just 2 issues with your correction:
> 
> 1) I don't believe your comment (or the RFCs you cite) reflect currently deployed reality.
> I've seen the lines of code - they allocate from 32 dynamic types, not 192.
> So the solution you propose is going to be incompatible with deployed code bases.
> 
> 2) If the true limit at which one has to change allocation strategy were to become 96, not 32, it actually strengthens my "falling off a cliff" argument (unless you don't believe in applications with more than 96 streams): This won't be a problem except in rare cases - ensuring even more code will be deployed that works until stressed.
> 
> I also have a confession to make: On some days, I respond to mail addressed to me without checking that I'm caught up on the mailing list that the conversation is copied to.
> 
> In this case, I responded to the mail sent directly to me on May 12 (a Sunday) without checking that I had also read your posting to the list on May 8 (Wednesday on the week before).
> 
> I suppose that's enough of a lapse in courtesy to be deserving of some censure.
> 
>          Harald
> _______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb