" it makes sense to always try these ports even if the current device has no 
6LoWPAN interfaces."

On a server I probably agree.  I doubt shortage of these sockets on a server.

On a client we should only ask for them if we expect to benefit.  But how do we 
tell?  I'd be willing to take the application's word for it (bit in flags), but 
I wouldn't want to do it by default.

John Light
Intel OTC OIC Development

-----Original Message-----
From: Macieira, Thiago 
Sent: Wednesday, June 24, 2015 11:46 AM
To: Light, John J
Cc: Carsten Bormann; iotivity-dev at lists.iotivity.org
Subject: Re: [dev] Proposal for IP Adapter and request for feedback

On Wednesday 24 June 2015 11:33:41 Light, John J wrote:
> I don't know how header compression works.  At what point does the
> referenced UDP LOWPAN_NHC Format come into play?   Are both ports created
> first, so you know what you've got to work with?  Or do you only know 
> one port at a time and have to make assumptions about the other port?

> If the ports are created first and you pick the Format after, then 
> pulling from a 256 entry pool seems doable.

> One concern is tying up limited space without actually using header 
> compression.

Header compression is applied by the node that transmits on the 6LoWPAN 
adapter. So it makes sense to always try these ports even if the current device 
has no 6LoWPAN interfaces.

But like I said, it might be best to come up with a Linux kernel feature to 
suggest these ports and let it do the finding of the free one available.
--
Thiago Macieira - thiago.macieira (AT) intel.com
  Software Architect - Intel Open Source Technology Center

Reply via email to