[Olsr-users] P2P chat

[Juliusz Chroboczek]

> The idea about our "debug chat plugin" was that you can somehow reach
> the other admins of the local nodes in case of a mayor routing problem.

Ah, I see.

We're speaking about two different things here.  Mitar and I are
thinking of a general purpose chat/PA system that happens to run over
the mesh, while the three of you are thinking of an emergency PA system
for when the network has broken down.

Now if I were to design the latter, I'd still not bind it with the
routing daemon.  Instead, I'd build my own flooding mechanism into the
chat software, so that it can continue to work even when the routing
daemon has crashed.

Note that this is exactly what ahcpd does: it implements its own,
relatively efficient, naive multicast transport using link-local IPv6
multicast only.  (It could equally well have been done using v4
broadcast or even link-layer broadcast, but IPv6 link-local is a little
bit cleaner than either of the alternatives.)  FWIW, the multicast
transport implementation is just 230 lines of code (file transport.c in
the ahcpd sources); the remaining 4000 lines of ahcpd is higher-level
stuff.

                                        Juliusz