Note to whoever owns the list: Set the Reply-To to the list please :) On Tue, 2 Dec 2003, Andrew Miller wrote: > The original mail suggested implementing this as a channel mode. It would need > to propagate to other servers, and clients(at least chanops) in the channel > would need to be informed that the mode had been set. It could, of course, be > implemented with a new command in a similar way to TOPIC. Of course, it would > then have to propagate at the burst and we could have the same old TOPIC > bandwidth issue again. It would need a way to be unset as well.
Yes, propagation would be an issue. Mode +k would need sending separately from other modes to prevent misinterpretation (+kl blah 12). Unsetting is hardly a problem, a simple resetting of the key without the text would accomplish it. > As for notifying clients trying to join the channel, probably a numeric would be > the most useful. We already have one. As I said, a well placed %s would add this feature. > If we are going to implement this, it could expire after a network-wide > timeperiod. For example, channel DoS attacks are usually transient situations, > and expiring the "Turing test" question after 30 minutes might not be a problem. I don't see why you would force a duration. > Alternatively, the channel could be set +r, as the cservice set has a similar > check on it anyway. This does however impose a significant time burden on people > wanting to join the channel, and may simply result in people using another > channel, and fragmentation of channels. Except, presumably, for the army of drones who are logged in with X precisely to circumvent cmode +r. +r unfortunately isn't really an option for most channels because a) people won't bother logging in and the channel base fragments as you said, and b) the effectiveness of it against a determined attacker is practically zero. GK PS: I realise the effectiveness of any security precaution against a determined attacker is practically zero. This doesn't mean trying is a bad thing.