Proposed alfs communication model

Matthew Burgess matthew at
Wed Feb 16 13:15:19 PST 2005

Jamie Bennett wrote:
> I've put up a little proposed communication model for alfs the way _I'd_
> like to see it done at 
> So a typical message sent from client to server on a different
> machine would be :-
<snipped diagram>

These two diagrams aren't quite the same.  Your ASCII art didn't show 
the dbus interaction at all.  For the sake of my own sanity I'm going to 
write things down in a sequential order of how I think things will work 
- please let me know if I misunderstood anything:

1. alfs-ui sends a plain text message to client dbus
2. client comms module listens out for and picks up the message from the 
client dbus
3. The plain text message has some indicator that tells the comms module 
where to pass the message on to (e.g. "server-destination: localhost" or 
"server-destination:").  The client comms module 
SOAPs the message and passes it to the server comms module
5. server comms module unSOAPs the message and sends the now plain text 
message on to the server dbus
6. alfsd picks the plain text message off of the server dbus.

In the situation where alfs-ui and alfsd are on the same box we get:

1. alfs-ui sends a plain text message to dbus
2. alfsd picks up message from dbus.

So, it looks like all 3 elements (alfs-ui, comms module and alfsd) need 
to be able to listen and talk to the dbus.  I therefore presume the 
messages would all have some indicator of not just the IP address of the 
intended recipient, but also the specific module that needs to pick the 
message up and act on it?  Or is it a case of each module in turn picks 
the message up and simply asks whether it knows how to act on that type 
of message?



More information about the alfs-discuss mailing list