[ax25-layer2] 7-byte address proposal

[Samuel A. Falvo II]

On 8/2/06, scott at opentrac.org <scott at opentrac.org> wrote:
> If you want to avoid interference with existing AX.25 traffic, why not use a
> different FCS polynomial?  Or simply take the one's compliment of the FCS
> and you can be sure that it'll never appear to be a valid frame.

I would have to say because the FCS is already standardized, and
indeed, it works perfectly well.  Additionally, many hardware TNCs
have dedicated hardware to help handle HDLC frames -- if we use
different FCS algorithms, it'll break these TNCs, even in KISS mode.

Finally, the HDLC specification, as I understand it as I don't have
the original specs, states that the address field is variable length
-- it is *only* AX.25 2.2 and earlier that says that the address field
must be an integral number of 7-octet addresses.  HDLC in and of
itself couldn't care less -- as long as bit 0 of the last address
octet is set, that's all that matters.

Therefore, using a sentinel and AFI is fully compliant with the HDLC
specifications, and *should* be compatible with hardware HDLC
implementations as well, but not having the datasheets for those that
are commonly in use (what ARE commonly in use?), I cannot be sure.

The AFI approach is just a suggestion at this point.

-- 
Samuel A. Falvo II