[ax25-layer2] different perspective?

[Wes Johnston]

Comments below...

pete at ae5pl.net wrote:
> The problem with this, from a layer 2 standpoint, is that it puts the
> definition of the LAN coverage back on the endpoints, not on the network
> appliances (digipeaters) if they exist.  If they don't exist,

Perhaps it wasn't clear in my examples.... the first digi to repeat a
packet under my suggestion does a call swap just like NSR does.  Except
instead of swapping in it's own callsign, it swaps in a group name.  So
the end users still would have no control over their paths if they used
generic callsigns like WIDE... or anything for that matter.  If they
used specific callsigns then the digi would not repeat their packet
unless of course it's callsign matched.

> This can cause layer 3 protocols to essentially break because they will
> be trying to find devices that might appear "local" but be many hops
> away.  The reason that the digipeaters using the NSR algorithm enforce

I could see that some well meaning digi owners might start to create
groups that are too large... I can see some folks not knowning any
better creating a group that is, say, eastern USA.

> path restrictions is two-fold: allow the digipeater operator to control
> what their digipeater repeats (reduction of QRM) and to provide a
> relatively static LAN configuration for discovery processes.  This last

The static configuration is what bothers me.  Of course net/rom's
dynamic tables were slow to expire stations heard in band openings were
also bad... there has got to be middle ground w/o too much overhead of
sending routing data over the air.  Do we think it would be possible to
incorporation RSSI readings into a routing table so the TNC knows it's
closest (ie strongest) neighbors?

> Your perspective, as you stated earlier in your post, is trying to use
> AX.25 as a layer 3 protocol.  The problem with this is the same problem
> we have run into before: this creates a unique "protocol" which is not

Umm.... I was just keeping an open mind for how layer three would pass
packets on to level 2.

Wes