Fwd: Re: [aprssig] Packet routing, path specification.

[AE5PL Lists]

> -----Original Message-----
> From: Robert Bruninga
> Posted At: Thursday, June 23, 2005 9:59 AM
> Subject: Re: Fwd: Re: [aprssig] Packet routing, path specification.
> 
> I do think that NSR can co-exist as long as it permits sender 
> chosen discrete paths as needed.  What I completely object to 
> is it's draconian stated approach that will prevent ANY 
> source routing under any
> conditions.   Anyone who understands paths
> will know that a directed path is MUCH MORE efficient than 
> ANY broadcast protocol to the same distance.  

Anyone that understands networks knows that a directed path (source
routing) of any form breaks the network (proven over and over again in
AX.25).  And allowing any source routing into the algorithm breaks the
network (yes, it is already broken; NSR is trying to fix it).

Your example of wanting to check into the Philly APRS net on Sunday's.
Why is it so important for you to do this via RF?  If you send a message
(that is what checking into a net is, right?) to the net control, your
message and your position will be gated to the Philly LAN.  Gee, no need
for discrete paths.  That is the key for getting rid of source routing
COMPLETELY!  APRS has _no_ demonstrable need for source routing (at
least you have yet to provide one example other than "because I want
to").  There is no such thing as "ROUTINE and IMMEDIATE" requirements
for getting your non-message packets to a distant area on the user's
say-so.

Bottom line: As I stated in the beginning, those that will feel the
"pain" of NSR are those that insist on trying to force long distance
communications through a local area network.  What those users will find
is that there is magic in gateways that get messages anywhere world-wide
AND the associated posits (gee, I can have human-to-human communication
anywhere in the world with no knowledge of the network that carries the
packets.  What a novel concept!).

The NSR algorithm purposely ignores the content of a packet as it is an
AX.25 UI digipeater algorithm, _not_ an APRS algorithm.  It is designed
to efficiently move UI packets about a local area with gateways
providing long distance communications links.  And that is what APRS
really does for people (a tactical protocol, right Bob?).  If we tried
to put everybody world-wide on each RF network, the RF would be useless
and therefore APRS would be useless.

73,

Pete Loveall AE5PL
mailto:pete at ae5pl.net