Re: [GNUnet-developers] gnunet-chat

[Igor Wronsky]

On Mon, 16 Sep 2002, Christian Grothoff wrote:

> Also, it sends n^2 messages (worst case) per message - to all peers, 
> regardless if people are subscribed or not. So this is really JUST and only a 
> demo-application. It's not *meant* to be for production use, ever. It's meant 
> to be educational code.

Note that people might not heed such considerations. ;) Typically
anything will be used for a purpose if there is no better alternative,
no matter how bad or harmful the current one is. 

I suggest, in capital letters, a note on top of gnunet-chat man page
"do not use this software - for study purposes only", and perhaps 
on startup as well. :) Even a safer solution would be not to 
install the whole thing at all unless the user is smart enough 
to request it explicitly and understand all the implications.

But if a chat could be implemented in a little less bandwidth-
consuming way and more securely, it would of course be quite
useful. From what I understand, is the node operator now
able, by gnunet.conf, to specify what kind of traffic the
node is participating in? e.g. chat messages will not be
forwarded/received if "chat" is not on? (same goes for future
apps like 'video conferencing') 

> Every chat message that gnunetd receives is broadcasted to all connected 
> peers. The peers check if they have seen the message already, and if yes drop 
> it. If not, they send it to all connected gnunet-chat clients and again 
> broadcast it to all of their connected neighbours. 

I wonder whether gnunet-chat will make the nodes millionaires
or beggars overnight ... ;) Hmm. Perhaps not. The one who talks
is paying, I imagine, and talks not more than he can pay for,
unless there is free bandwidth. Ok.


I.