Re: [Manet-dt] Re: PacketBB

On Mar 28, 2007, at 5:45 PM, Templin, Fred L wrote:

>> My goal would be to save some substantial packet size in the average
>> case.
>
> But can you do this without loss of generality?
>
> FWIW, I advocated a TLV approach for packet encoding a
> long time ago but was beaten back by forces that preferred
> efficiency above all. If you're not careful, what you end
> up with is a proprietary solution and/or limited domain of
> applicability.

I concur with what Fred Templin states above. OSPF went down the same  
route with their LLS...

A lot of efforts were spent on the design of packetbb - in  
particular, in achieving a general format offering extensibility,  
parse-ability and compactness -- the latter through compressable/ 
aggregateable addresses and modular message/packet/tlv headers. And,  
I may add, in an open process where all input that was brought  
forward was carefully reflected on and taken into account as best  
possible. Chris has already cited examples hereof, so I won't do that  
unless prodded to do so.

I'll not reiterate Chris' email, but I agree with the general  
comments he make, that there're requirements pulling in opposite  
directions for the design of packetbb. My point of view is, that  
packetbb offers a good compromise between these requirements. I can  
only speak for OLSR, however message sizes are in the "average case"  
for OLSRv2 expected (and observed) to be less than for OLSRv1 -- with  
the added benefit of a more flexible/extensible format to boost.

I'd really not like to revert to a proprietary packet/message format,  
which will leave us with no extensibility and complicated parsing  
logic. I'd also rather not have to spend another 2+ years in  
development of an alternative format, when one exists which I find to  
be a good compromise.

Thomas

-- 
Thomas Clausen
Thomas.Clausen@polytechnique.fr
http://www.lix.polytechnique.fr/Labo/Thomas.Clausen/
http://www.lix.polytechnique.fr/hipercom/

_______________________________________________
Manet-dt mailing list
Manet-dt@ietf.org
https://www1.ietf.org/mailman/listinfo/manet-dt