[aprssig] Precedence Bit and IP Encapsulation

[Iain R. Learmonth]

Hi Jason,

On 19/05/2020 17:12, Jason KG4WSV wrote:
> That seems a pretty small win in exchange for requiring development of a
> completely separate transport layer when there are a couple of pretty
> good ones already sitting there.

Over a dial-up link with PPP it might be a big win. A UDP header is
almost 20% the size of a large APRS frame (64 bytes vs 330 bytes). If
there's a smaller APRS frame then the UDP header makes up an even
greater proportion.

> Is this the same standard that's falling out of modern kernels for lack
> of development/modernization?

In Linux, the situation looks bad but is limping along. Many AX.25 BBS
link together using AXIP or AXUDP, preferring AXIP where possible. BPQ32
support AXIP and AXUDP.

> What's the use-case for an ax25/ip network?

To create a virtual AX.25 point to point link.

The two cases where I'm planning to use it for HamBSD will be:

(1) to allow HamBSD development on a computer that isn't directly
connected to the TNC. This is currently possible with KISS-over-TCP to a
limited extent but requires something like socat to bridge a TCP port to
a serial port and does not provide any security.

(2) to provide fallback links between APRS infrastructure sites. We have
an interesting situation at the moment where a digipeater can hear the
IGate but the IGate doesn't hear the digipeater. If packets can get from
the digi to the IGate via the Internet, that means that 2 way
communication can be established. It's not possible to transmit packets
from the tunnel out of the digipeater due to license restrictions, but
that's OK because the RF path works in that case.

I imagine there are also other use cases that I won't try out myself but
would be supported.

> Sorry, i don't mean to be argumentative (as I probably sound), I just
> don't understand where you're going.

Questions are good.

Thanks,
Iain.

-- 
https://hambsd.org/