Re: Re the encapsulation I-Ds...

[Dr G Fairhurst <gorry@erg.abdn.ac.uk>]

Thanks John....

You could have mailed the list :-)

John Border wrote:
> 
> 
> http://www.ietf.org/internet-drafts/draft-fair-ipdvb-ule-00.txt
> http://www.ietf.org/internet-drafts/draft-clausen-ipdvb-enc-01.txt
> 
>     A couple of comments which apply to both documents...
> 
>     ENH - Re Section 5.2, Last Paragraph
> 
>         In most cases, latency will be more important.  But, in some cases, a
> trade-off for more efficiency might be desirable.  So, how about having the
> encapsulator optionally support waiting N milliseconds for another SNDU, where
> N is configurable with a default value of 0?
> 

Yes, that seems OK. 

But does it *really* matter if N is some small value (e.g. value <30 ms
or so ?).   

>     IMP - General
> 
>         As far as I can tell, there is no way a receiver can automatically
> detect which encapsulation method is being used by the encapsulator (between
> these two methods or between one of these methods and the DVB-S standard
> mechanism(s)).  Therefore, I am assuming the receiver knows a priori via
> configuration (or some other means, independent from the encapsulation
> itself), probably associates with the PID being used.  Of course, I may have
> just missed the subtleties which allow automatic detection of the method.  My
> real point is that a sentence or two in the introductory material re how the
> receiver knows which encapsulation method is being used should be included in
> the documents.
> 


I separated the two scehems to help the list think through things. 

Basically, the LE (draft-clausen-ipdvb-enc-01) scheme
requires and uses the PES semantics for the MPEG-2 PUSI bit and requires the
use of the MPEG-2 TS Adaption Field. This only happens for certain types 
of MPEG-2 Transport Streams.

The ULE approach assumes you will not supply an Adaption Field, and that
every time the
PUSI is set, the first byte of the MPEG-2 TS Packet contains a 1 B pointer.

I suggest if we need to differentiate the "modes" we could do that as a part
of "address resolution" - i.e., when the receiver figures out the PID, it
should also associate the PID with MPE, LE, or ULE! (or whatever subset
survives the next round of debate). 

Gorry

> John Border
> Hughes Network Systems, Inc..