Hi Cees,
Michaël and I work on a project where we need to support discovery of hosts in a given network. We spent some time looking at AardWorks code and applications (Chat, and Kolibris). But, we aren't sure about how users learn about each other.
Imagine that x don't now anybody. Then how x can learn about existing other peers ? My question is about how to "bootstrap" the system. We understood the idea of Gossiping. But, how the first peer is discovered in AardWorks applications?
Thanks, Noury -------------------------------------------------------------- Dr. Noury Bouraqadi - Enseignant/Chercheur Ecole des Mines de Douai - Dept. G.I.P http://csl.ensm-douai.fr/noury
European Smalltalk Users Group Board http://www.esug.org
Squeak: an Open Source Smalltalk http://www.squeak.org --------------------------------------------------------------
Noury Bouraqadi wrote:
But, how the first peer is discovered in AardWorks applications?
It'd be really cool if Squeak supported Bonjour (nee Rendezvous): http://www.apple.com/macosx/features/bonjour/
:-)
Tony
Noury Bouraqadi wrote:
But, how the first peer is discovered in AardWorks applications?
There are two easy possibilities: - Multicast for LAN-based discovery, well-documented in the Jini specs; - Nodes collaborate to put bootstrap lists into well-known places (used by Gnutella, IIRC).. AardWorks Gossip, being mostly internet-oriented, currently uses the latter one - every x seconds the list of known peers is posted to a well-known URL, and a node that doesn't know about any peers can fetch that list to seed its data.
Adding Multicast-based discovery would be relatively easy, though.
squeak-dev@lists.squeakfoundation.org
-
Cees De Groot -
Noury Bouraqadi -
Tony Garnock-Jones