Hey Sam- Not checking the list as often as I should, so sorry for the delay. Great you guys created a separate module for the RUDP stuff -- not many (any?) of those out there in Java, so that'll be huge.
The STUN stuff doesn't do any reliability. STUN is actually really straightforward -- STUN clients just query STUN servers for their public addresses using a binary protocol. It actually doesn't even do any NAT traversal stuff -- that's all really handling in ICE. The "ST" in the name for "Simple Traversal" is actually changed in the latest drafts to alleviate the immense confusion that caused. So, basically STUN could be used along with something like SIP and ICE to negotiate a connection over which LimeWire's RUDP layer could be used. -Adam On 3/14/07, Sam Berlin <[EMAIL PROTECTED]> wrote:
A month or so ago some of the FreePastry folks contacted us at LimeWire about our RUDP implementation to see how they could use it, which led to us decoupling the RUDP implementation from the rest of the code. Right now it's a separate component called 'rudp' (and depends on a few other LW components: 'nio', 'common', etc..), but it has a pluggable message implementation. That means it'd be relatively trivial to implement messages that conform to STUN instead of Gnutella. The address discovery and NAT punching are left up to the implementor to plugin -- the component is purely for a reliable UDP transfer. I'm not as up-to-date on STUN as I should be, but is it intended for a reliable UDP stream also? If so, it would be a really neat/fun project to work on. Sam On 3/14/07, Adam Fisk <[EMAIL PROTECTED]> wrote: > Just a quick note that the LimeWire address discovery code only relates to > STUN in the loosest sense possible. They both discover your public address, > but that's about where the similarities end. The wire protocols are > completely different, with LimeWire using a series of Gnutella messages and > with STUN using, well, STUN. > > -Adam > > > > On 3/13/07, Jeff Hoye <[EMAIL PROTECTED]> wrote: > > Gwendal, > > This may be waiting on me for just a bit longer (~1 month), as I'm > trying to reorganize the transport layer to be modular so > > that such an integration will be almost trivial. I hope to have an > interface that just asks for a socket or to send a datagram to > > an IP:port, and the corresponding callbacks. This will make it much > easier than searching the existing codebase for the places > > where sockets are opened or datagrams are sent. > > I know of some folks on our mailing list who are looking into > integration of Limewire's STUN impl. However, because of our > > differing licenses (Limewire is GPL, FreePastry is BSD-like) the > distribution may be a bit tricky. I imagine we'll figure out > > those details when the time comes. Perhaps someone can start a > SourceForge project for that. > > > > -Jeff > > > > SIMON Gwendal wrote: > > > On 3/13/07, *Jeff Hoye* <[EMAIL PROTECTED] > > > <mailto:[EMAIL PROTECTED] >> wrote: > > > > > > NAT support is currently limited to users who can set up port > > > forwarding, or have UPnP, but some of our users are investigating > > > integration STUN to expand this capability. > > > > > > > > > For student internships, we also consider some devs on networking issues > > > related to DHT, especially NAT traversal and IPv6. We would be pleased > > > to contribute. Is there any coordinated action on this topics in the > > > FreePastry community ? > > > > > > Cheers, > > > > > > -- Gwendal > > > http://enstb.org/~gsimon > > > > > > > > > > ------------------------------------------------------------------------ > > > > > > _______________________________________________ > > > p2p-hackers mailing list > > > [email protected] > > > http://lists.zooko.com/mailman/listinfo/p2p-hackers > > _______________________________________________ > > p2p-hackers mailing list > > [email protected] > > http://lists.zooko.com/mailman/listinfo/p2p-hackers > > > > > _______________________________________________ > p2p-hackers mailing list > [email protected] > http://lists.zooko.com/mailman/listinfo/p2p-hackers > > _______________________________________________ p2p-hackers mailing list [email protected] http://lists.zooko.com/mailman/listinfo/p2p-hackers
_______________________________________________ p2p-hackers mailing list [email protected] http://lists.zooko.com/mailman/listinfo/p2p-hackers
