[ct-user] serial nr sequencing - an other view
Zoli Pitman HA1AG
Zoli Pitman HA1AG" <firstname.lastname@example.org
Sat, 2 Jun 2001 08:50:37 +0200
I've been reading your debate about serial number sequencing with great
In my opinion a sort of client/server architecture would be the best
solution for the number sequencing problem. In the local club there's a
piece of self-made SW, which does it flawlessly in network.
The Run station is the one, which issues a number upon the request of a Mult
station. In practice when the Mult op finds a station he wants to work, he
presses the Grey+ and the Mult PC requests a number from the Run PC. The Run
PC inserts a blank record in its log immediately and requests QSO data from
the Mult PC. This is a point-to-point communication.
When the Run station makes a QSO, it broadcasts the QSO data to all the
Mults (serial included). This is a point-to multi point type of
communication without acknowledgement messages.
When a Mult PC crashes, no QSO can be lost except the very unlikely
situation if the crash happened between the Serial Nr Acknowledge message of
the Mult PC and the entering the QSO data in the Mult PC. (It never
happened). If the Run PC crashed, no serial number could be requested by the
mult stations, so no QSO can be lost either.
If two stations swaps roles (Run becomes Mult), the flags have to be changed
on both computers. Of course, only one Run (number server) can be present at
73 de Zoli HA1AG
Administrative requests: ct-user-REQUEST@contesting.com