SJM> I remember having a problem here in Madrid because I was
SJM> using a different netmask (Madrid 184.108.40.206/255.255.255.0)
SJM> to the netmask used by another station (ampr.org
SJM> 220.127.116.11/255.0.0.0) on the same frequency.
SJM> If the interfaces of the 2 hosts use different netmasks,
SJM> they will not recognise the broadcast request as such, and
SJM> they will thus get ignored. Correct?
The broadcast address can technically be differentiated from the netmask,
although ordinarily the broadcast address is
interface-address | ~netmask
SJM> However, I've always wondered whether I should be using the
SJM> real ampr.org netmask for *all* radio ports, something which
SJM> I think won't work well with routing. If people sub-net, as
SJM> we do in Madrid, then someone using the general netmask for
SJM> 18.104.22.168 will find problems and may not be aware of the
SJM> Could the problem being reported be a simple symptom of
SJM> wrong netmasks, something which won't be obvious, unless you
SJM> see the broadcast requests from another machine, and check
SJM> the broadcast address corresponds with that configured for
SJM> the local host's own radio interface?
The only critical thing is that everyone on a LAN -- that is, who hear each
other directly -- must be using the same broadcast address on that LAN.
It is certainly wrong to configure an interface with netmask such that it
thinks all 44.x.x.x hosts are in its local LAN. You would never want a host in
Madrid to send an ARP request for a host in California on 144 MHz. It is up to
local consensus to define the appropriate netmask and broadcast address for a
LAN, and everyone needs to use the same values. Once this has been
accomplished, a separate route for hosts within Amprnet (44.x.x.x) but not in
the LAN can be configured, since it will need a gateway within the LAN.