rohc team mailing list archive

[Didier Barvaux <didier@xxxxxxxxxxx>]

Hello,

> Sorry for early alert, I found the answers from previous mails. But
> now lets ask difference between R-Mode and O-Mode.
> 
> As I read from rfc 3095 R-Mode context updates is more robust then
> O-mode. Does context update mean sending IR type rohc packets ? Or
> sending newly created context information to decompressor which is
> done once?
>
> Are there any other differences between O-mode and R-mode?

In both modes, context updates are done with IR or IR-DYN packets.
Those IR/IR-DYN packets are repeated several times until the compressor
gains confidence that the decompressor got the update.

The R and O modes vary on the way to gain that confidence:
 * In R-Mode, the compressor gains confidence if it receives an
   acknowledgement of at least one of those IR/IR-DYN packets from the
   decompressor.
 * In O-Mode, the compressor gains confidence in two ways, the first
   one to occurs is enough:
    * if it receives an acknowledgement of at least one of those
      IR/IR-DYN packets from the decompressor.
    * if it sends N IR/IR-DYN packets in a row (N is not defined in
      the standard, but left to implementers).

The threshold N is not yet configurable in the ROHC library. It is
defined to 3. It is planned to change that in the near future.


> As read from documentation r-mode is still not implemented. Is that
> true? Is there any plan to implement in the future?

R-Mode is not yet implemented in the ROHC library. There is plan to
implement it in the near future for the moment. If you are interested
in adding R-Mode support, I may help you however.

Regards,
Didier