[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Anti-collide patch




On Sun, 10 Jan 1999, Joern Westermann wrote:

> Michal *pht* Svoboda wrote:
> 
> > this makes me get another idea, if Guest 5432 changes nick to 2345, the
> > server A could then FNC his 2345 to 5432 (i.e. the nick of the collider) ?
> > 
> Let me explain the problem:
> 
> 3 servers, A - <big net> - B - C
> The connection between B and C is lagging in direction towards C (e.g. C
> has recently connected). There're lots of messages in the buffer from
> B->C, but the queue C->B is empty since C only has to distribute its own
> users as a leaf.
> So any message issued by C gets to B/the net/A a lot faster than any
> message from A to C.
> Assume a client "victim" on A and a collider "killer" on C: The "killer"
> can change to any nick it wants on C since all kill or nick messages
> from A/B haven't got through to it yet. There's no nick delay since
> there were no kills on C yet. But all "Nicks" are send to B with almost
> no delay and B generates a kill towards the net and C, if it detects a
> collide.
> 
> Your proposal of changing the victim's nick back to the nick of the
> collider would only work if there is some kind of local nick history
> which prevents to use nicks which were taken by any client no longer
> than 5-10 minutes ago. But then there's probably a 2nd lagging server...
> 
maybe i'm totally wrong but if clientA on server A will get collided by
nick change clientC->clientA on server C, and server A will change the
nick of it's clientA to clientC, what should go wrong?

pht