[Nice] Reliable vs. Unreliable...

Youness Alaoui youness.alaoui at collabora.co.uk
Wed Aug 25 12:17:46 PDT 2010


Once connectivity is established, a non-reliable agent will act like a UDP
socket.. a reliable agent will work as a TCP socket.
Yes, you should make sure buffers fit in one MTU, although the kernel will split
packets for you, if one chunk is lost the whole message is lost (just like UDP),
so it's better to split your packets yourself.
Also, with a non-reliable agent, just like UDP, one send, will result in one
recv on the other side, it's not a stream, but with a reliable agent, you could
send multiple packets and they'll be received in one shot on the other side,
since TCP has no notion of 'packet boundary'.

Youness.


On 08/25/2010 11:25 AM, Tony Di Croce wrote:
> If I don't create a reliable nice agent, are my calls to
> nice_agent_send() limited to buffers that can fit in one MTU? Or does
> libnice just send multiple packets?
> 
>    td
> 
> 
> 
> _______________________________________________
> Nice mailing list
> Nice at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nice


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 262 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freedesktop.org/archives/nice/attachments/20100825/2b2d9aa4/attachment.pgp>


More information about the Nice mailing list