Re: [rohc] Liaison towards 3GPP

["Lars-Erik Jonsson" <lars-erik.jonsson@ericsson.com>]

First of all, I think Magnus H. and Krister S. both made good comments. However,
I will say some final words about this violation issue.

>> Messing with the payload (RTP header is application data) is to violate
>> the
>> integrity of the packet. Do you think that is OK????? I do not!!
>>
>>
>>
>Me too, in some cases, but in some other cases this can be tolerated,
>since we want for instance to allow delivery of errored PDUs
>when voice is transported. I cant figure out anything
>messing around integrity more than errors...
>


You can not compare errors to actions taken to purposely "damage" the payload.
Even if you want damaged payloads to be delivered, you can not defend errors
introduced by the network on purpose. Usually, there are mechanisms to protect
against real errors (UDP Checksum) and this is a protection required by the
sending application. Even with UDP Lite, I can not see why the application would
choose not to protect the RTP header part.

I can see reasons for having non-IP-based solutions when you do not want the
advantages and flexibility of IP, but then it should be clear that we are
discussing two different things. To call a system "IP-based" even when using
things such as header stripping is like calling WAP a mobile Internet solution.

/L-E