Re: [dtn-interest] Comments on DTNRG documents from BBN SPINDLE project

From: Scott Burleigh <Scott.Burleigh(at)jpl.nasa.gov>
Date: Sun Apr 03 2005 - 19:24:57 EDT

Michael Demmer wrote:

>>- Agree that in some -- not all -- cases we may have multiple
>>"application instances" (we can probably find another term that is
>>more palatable; I'll say "clients" for now, though that may turn out
>>to be confusing) that are concurrently using the services offered by
>>a given bundle agent.  In those cases we need a way to mux/demux
>>bundle payloads from/to those clients, something functionally
>>analogous to port number.

>
>
> Ok as well. We probably need some concept that corresponds to the

You're right. If we use numbers to represent these things then I would propose that we use zero for this purpose.

  I could actually envision several uses for
> this facility, in that routing protocols could (and probably should)

>>- Because we don't route on these source and destination mux/demux
>>tokens, because they are optional, and because they are only
>>relevant at the point at which you've reassembled the payload from
>>fragments as necessary (assuming we're still doing bundle
>>fragmentation, which I will argue against at a later time), I would
>>suggest that we add them to the Payload header.  

>
>
> I'm tempted to agree, however this may cause problems for status

Matt brought up the same point, and I agree. I'd like to make them SDNVs, and yesterday I suggested that we put them in the Dictionary header. But that breaks if we're clever enough to omit the Dictionary header in some cases (as you mentioned a couple of emails ago). Maybe the best answer is to have them at the tail end of the Primary header, even though that makes the Primary header variable-length.

Scott

---

dtn-interest mailing list
dtn-interest@mailman.dtnrg.org
http://mailman.dtnrg.org/mailman/listinfo/dtn-interest Received on Sun Apr 3 19:35:47 2005