Sachli wrote:

>>Thanks for the confirmation.
>>I dropped Events when Borland messed them up in 5.x and have
>>my own messaging package that posts the record number that
>>has changed so that only interested slaves need to refresh
>>so I had not kept an eye on progress. ;)
>
> Interesting!
> Can you share it to us how you do the messaging?

It's just a separate module. Each machine has an application
called LocalHub which uses Multicast to communicate with
MasterHub. When an update is required ( a train leaves a
station for instance ) that is sent as a message from a
LocalHub to MasterHub, which relays it to all the other
LocalHub's (It's a frame relay thing - the MasterHub has the
wide bandwidth link :) ). If that train is going through the
station of a receiving LocalHub, the LocalHub updates that
record in a local copy of the database. Simple replication,
and so only required refreshes go over the wire, so the
whole thing works with just 64kBaud links at each station.
In the gaps each hour, LocalHub downloads another hours
worth of train movements and the whole thing just keeps
rolling.

All the local applications work into the local database, and
talk to LocalHub, which relays the messages. The activity to
LocalHub is also used to update the network status, so the
control centre knows what is going on. If a link is lost,
the local displays can be maintained, using a simple timed
depart, and the database just gets rebuilt when the
connection is restored.

(p.s. while we had to pay for licences the slaves were
simple dBase engines - with Firebird that ceased to be a
problem ;) )

--
Lester Caine
-----------------------------
L.S.Caine Electronic Services