Fuzzy Fox schrieb:
Dan Kegel [EMAIL PROTECTED] wrote:
I've been working on a way for peer-to-peer games to work through
NAT's transparantly, with some success.
http://www.alumni.caltech.edu/~dank/peer-nat.html
Your work seems to be a technique that game programmers can use, to make
their programs more NAT-friendly.
It's also a technique for NAT programmers to make their NAT's
more compatible with NAT-friendly games. Masq by default isn't
compatible at all with NAT-friendly peer-to-peer games; I give a
patch that shows how to make it compatible. (I plan to have
a better patch later.)
It doesn't seem to do anything for
those of us who want to play the games that are not NAT-friendly.
That's the type of audience you're writing to, on this list. :)
Well, there are so few people on the masq-dev list, I thought maybe
I'd reach more Masq developers by posting on the masq list, too.
I'm still amazed at how many games use bizarre protocols involving UDP
packets on multiple ports, to have their games talk to one another.
What's so wrong about opening a single TCP socket connection, and just
sending data to the other machines?
True, the schemes they use are a little wierd, and NAT-unfriendly.
But you're wrong to assume that being NAT-friendly is easy.
Have you tried establishing a peer-to-peer connection between two
machines behind different NAT's? It's tricky.
For example, NAT's often will refuse packets from outside hosts
that aren't replies to packets sent from inside.
Think about that for a second - it's fine when talking to a
server not behind a NAT, but it's a chicken-and-egg problem
when two peers behind two different NAT's want to talk!
Worse yet, Masq as currently implemented takes a stream of
packets all from the same machine and port, and turns them
into a stream of packets from *different ports*. This
can play havoc with peer-to-peer networking, and is not
standard NAT behavior.
The web page
http://www.alumni.caltech.edu/~dank/peer-nat.html
shows how to get around these problems.
If there are any Masq developers out there, I'd love feedback...
Thanks,
Dan
--
Speaking only for myself, not for my employer
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
For daily digest info, email [EMAIL PROTECTED]
