On Feb 8, 2012, at 3:38 AM, Miguel A. Garcia wrote: > Hi Cullen, > > Thanks for your review. See inline comments. > > On 07/02/2012 18:43, Cullen Jennings wrote: > >> Consider the case of a chat room that is receiving messages at a rate >> of 10 per second (perhaps bad-attitude during the IESG plenary) and >> also has some clients in the chat room that can only receiving >> messages at a maximum rate of 5 per second ( perhaps an iphone via VPN >> over 2G). Section 6.1 does not say what should happen in this case. > > I understand the problem description. The question is whether we want to make > the application layer aware of potential congestion and take actions on it. > > I don't know how the chat room application could in the first place detect > congestion. Any ideas of what to write in the draft are highly appreciated.
Note - I'd be OK with publishing the document with no change but I think some text would improve the draft. I think the most important part is just point out that this is a problem and applications need to have a strategy to deal with it. I would not prescribe a strategy that apps had to implement but instead have some non normative strategies that an application might use. One strategy is that is any client falls more than 10 seconds behind, that client get disconnected.
