On Tue, Jun 24, 2003 at 08:46:09PM +0200, E. Hietbrink wrote:
 > Also I have not yet looked into the RSA stuff yet.

I thought the RSA client verification used IP addresses as part of the 
exchange, in order to prevent man in the middle "attacks".

> The ideal solution I would like to have implemented is every client using
> one and the same UDP port, namely 2592. That portnumber is no nicely
> reserved for netrek. Clients should send PDU's with an additional 4 byte,
> server assigned ID which the server uses to distribute stuff among daemonII
> processes. Relying on source ip address won't be sufficient in all cases.

ipv6 is a far better solution than NAT.  Why not support build PPTP into
the server and the client and just run GRE?  ;-)

The Netrek protocol was designed in the days when ipv4 was sufficient.

> I know it will be hard and there will probably be pitfalls, but i'm bored
> and like to try. So I ask you vanilla veterans: is there any reason I
> should not do this?

No reasons in particular.  I'll merge anything with the main code base if
it is useful and doesn't hurt comprehension of the code.

Of course, I'd be much more enthusiastic if you were doing client side GUI
usability and acceptability (eye candy) improvements.

James Cameron    mailto:quozl at us.netrek.org     http://quozl.netrek.org/

vanilla-devel mailing list
vanilla-devel at us.netrek.org