RE: [rohc] Update of ROHCv2 profiles draft

Kristofer,

Thanks for the reply. Pls see some comments below.

> -----Original Message-----
> From: Kristofer Sandlund (LU/EAB)
[mailto:kristofer.sandlund@ericsson.com]
> Sent: Monday, June 11, 2007 11:44 PM
> To: Kapoor, Rohit; rohc@ietf.org
> Cc: Ghyslain Pelletier (LU/EAB)
> Subject: RE: [rohc] Update of ROHCv2 profiles draft
> 
> Hi Rohit,
> 
> thanks for your comments. My answers inline.
> 
> Kapoor, Rohit <mailto:rkapoor@qualcomm.com> wrote on den 12 juni 2007
> 04:02 :
> 
> > Kristofer, Ghyslain,
> >
> > Thanks for the updated draft. Pls find few comments below:
> >
> > 1) Regarding use of timer-based compression in pt_1_seq_ts, I had
> > brought up this issue earlier, but haven't seen any
> > resolution for it in
> > the current draft.
> 
> Since we hadn't decided what to do on that last time we discussed it,
I
> left the following TODO in the draft:
> 
> // TODO:
> // - PT_0/PT_1 packets? Use "tbc-optimized" or "msn-optimized"
> 
> So it was not forgotten, we just preferred to get out a new revision
of
> the
> draft as soon as we could and let us continue the discussion later.
> 
> I will give a more detailed answer on this issue later this week or
next
> week

<Rohit> OK, look forward to your answer.

> 
> > 3) ts_stride_value and time_stride_value are defined as:
> >
> >          ts_stride_value: This parameter is set to a
> > user-selected value
> >          that is the expected increase in the RTP Timestamp between
> >          consecutive sequence numbers.  See also Section 6.6.8.
> >
> >          time_stride_value: This parameter is set to a user-selected
> >          value that is the expected inter-arrival time between
> >          consecutive packets for this flow.  See also Section 6.6.9.
> >
> > Are these intended to be user-selected values?
> 
> Yes, they are "user-selected" the same way they are that in 3095 in
the
> sense that they can be set to any value independent of what is in the
> packet
> stream (of course, there are just a couple of values that would make
for
> efficient compression). But maybe the word "user" is not the best word
> to
> use here, since it might be interpreted by some as if it is set up
> during
> configuration while it is *usually* set by inspecting the flow over
> time.

<Rohit> How about using RFC 3095-like language?

      TS_STRIDE: Predicted increment/decrement of the RTP Timestamp
      field when it changes.

      TIME_STRIDE: Predicted time interval in milliseconds between
      changes in the RTP Timestamp.

Thanks,
Rohit

_______________________________________________
Rohc mailing list
Rohc@ietf.org
https://www1.ietf.org/mailman/listinfo/rohc