Re: [RFC PATCH net-next v0.1 1/1] net: add GeoNetworking protocol

When considering our 700m long train example, I think for sure. Because
the range of 802.11p is roughly 500m, there can be a optimal inter-
mediary, which is in range of the device at the start of the train but
not in range of the device of the end of the train.

Many years ago i did some networking work on the Swiss local commuter
trains, think the Zurich S lines. They had Ethernet running the full
length of the train.

This then leads me to thinking about having two nodes, one on each
end, with an Ethernet link between them. Ethernet, or any cable based
medium, has different semantics. A broadcast is received by all. There
is no point rebroadcasting on the same interface.

So i think this needs to be thought about in the overall picture.

quoted

Does the standard say anything about multi-interface devices?

The standard is mostly written with cars in mind, so I don't think so.
After all, there is not even an ITS-Type for trains, just one for trams

So maybe for the vehicle side of things, interface to interface
routing is not so important, until somebody pushes forward with train
use cases. But what about the fixed infrastructure? How often do you
see nodes along some sort of linear structure, connected together by
cable? Or a row of blind corners which is known for accidents, you
could put fixed infrastructure along the way and cable them together,
so the vehicles can "see" around the corners?

Sorry for all these questions, but i want to make sure we have an
architecture which works for the most number of scenarios, and is not
going to be a dead end for more complex setups.

    Andrew