[ENet-discuss] ENet and NAT hole punching

[Jay Sprenkle]

Thanks very much for the response.
Why five packets? A few extra to insure delivery?

On Thu, Jan 20, 2011 at 7:43 PM, Philip Bennefall <philip at blastbay.com>wrote:

>  Hello there,
>
> In my implementation I simply open a UDP socket directly, send about 5
> packets with a small interval between each, and close the socket again.
> Then, ENet kicks in and attempts a connection. The ENet socket wrapper
> functions are more convenient, but obviously the result will be exactly the
> same so it just depends on your taste.
>
>
> Kind regards,
>
> Philip Bennefall
> ----- Original Message -----
> *From:* Jay Sprenkle <jsprenkle at gmail.com>
> *To:* Discussion of the ENet library <enet-discuss at cubik.org>
> *Sent:* Friday, January 21, 2011 2:36 AM
> *Subject:* [ENet-discuss] ENet and NAT hole punching
>
> Good evening,
>
> I'm interested in making my ENet powered application able to do NAT hole
> punching.
>
> After looking through the NAT hole punching RFC it looks fairly simple
> (section 2.3 here<http://pdos.csail.mit.edu/%7Ebaford/nat/draft-ford-natp2p-00.txt>
> ).
> What is needed is a way to send a single packet to a specific address and
> port number.
> The content of the packet isn't important since it will be discarded
> anyway. It's just used to get the NAT to remember the address.
>
> If I read the source correctly I could simply open a connection and let it
> fail.
> Is there any way to send this packet without going through all the
> overhead?
>
> Perhaps just call enet_socket_send() directly?
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cubik.org/pipermail/enet-discuss/attachments/20110120/c1b88286/attachment.html>