WowAce

Heh. I wrote this at the behest of some guildmates, then forgot about it. I'll try to pay more attention to it. :D

Regarding the protocol, I'd love to design a nice extensible protocol. The current one is pretty brittle, very admittedly, since it doesn't use any kind of key system - just a plain attribute queue. It wouldn't be hard to basically just use something like the HTTP query string spec: key=val&key=val&key=val, etc.

I very heavily trust ChatThrottleLib. I used it for Threat-2.0, and barring a few edge cases (Unthrottled AOE in Mount Hyjal, *cough*), it handles data marvelously. It does yield to impolite addons (for example, Threat-2.0 used to "stall out" when addons like DoTimer were consuming all the available bandwidth, but it's better than a disconnect), and it chunks messages too large to send otherwise, more or less making data transmission a no-brainer.

While a whisper comm solution is O(n^2), the actual amount of data transferred is miniscule compared to a lot of other addons. In terms of server load, a whisper comms solution is no worse than a shared channel solution; in practicality, it's likely even more efficient. Given X people in a channel, every message consumes (x * n) bytes + overhead; whisper comms consume 2n bytes + overhead, but unless a user comes in contact with more than half the people in the channel in-game in a given session, whisper comms use less bandwidth.

In the grand scheme of things, I don't think that RP mods could consume an amount of bandwidth significant enough to cause strain on client or server, and I say that having written mods that do. :P