RE: [rohc] draft-ietf-rohc-rtp-06.txt

[zhigang.liu@nokia.com]

Hi Mikael,

I completely agree with your argument below. However, it is 
not exactly what section 5.4.2.2 (feedback logic in O-mode) says. 
There, an IR packet triggers an ACK (this is not optional, right?),
which implies that the static header fields (as part 
of the context) are given higher weight than others. Those context 
variables unlikely to change during a session (e.g. RND)
are similar to the static header fields and therefore, should be 
treated similarly. In short, it is the logic in 5.4.2.2 that 
confused me and let to my current thinking.

As long as the logic is consistent, I have no problem with
completely relying on repetition in O-mode (if it is your
belief, though I have second thought on that).

Br,
Zhigang

> It all boils down to this: 
> In O-mode we believe that repetition L times is enough to 
> establish a set of context variables (X). 
> Why should we then use a different way to establish another 
> set of context variables (Y), when the consequences of not 
> establishing a variable from Y are the *same* as the 
> consequences of not establishing a variable from X?
> 
> Mikael Degermark