This subdirectory illustrates a number of the ACE ASX framework features using an ACE_Stream application called the Event Server. For more information on the design and use of the ACE ASX framework please see http://www.dre.vanderbilt.edu/~schmidt/PDF/C++-USENIX-94.pdf and http://www.dre.vanderbilt.edu/~schmidt/PDF/ACE-concurrency.pdf. For more information on the Event Server, please see http://www.dre.vanderbilt.edu/~schmidt/PDF/DSEJ-94.pdf The Event Server example works as follows: 1. When the ./Event_Server/event_server executable is run it creates two SOCK_Acceptors, which listen for and accept incoming connections from Consumers and Suppliers. 2. The ./Event_Server/Transceiver/transceiver application plays the role of either a Consumer or a Supplier (with the current implementation it can only play one role at a time). The transceiver process can be started multiple times. Each call should be either: # Consumer % transceiver -p 10002 -h hostname -C or # Supplier % transceiver -p 10003 -h hostname -S where 10002 and 10003 are the default Consumer listening port and the Supplier listening port, respectively, on the event server, "hostname" is the name of the machine the event_server is running, and -C and -S indicate that the transceiver plays the role of a Consumer or Supplier, respectively. I typically run the Consumer(s) and Supplier(s) in different windows to make it easier to understand the output. 3. Once the Consumer(s) and Supplier(s) are connected, you can type data from any Supplier window. This data will be routed through the Modules/Tasks in the event_server's Stream and be forwarded to the Consumer(s). Since the transceivers are full-duplex you can also send messages from the Consumer(s) to Supplier(s). However, the Event Server will warn you about this since it's not really kosher to have Consumers sending to Suppliers... 4. When you want to shut down the tranceivers or event server just type ^C (which generates a SIGINT) or type any input in the window running the Event Server application. What makes this example particularly interesting is that once you've got the hang of the ASX Streams architecture, you can "push" new filtering Modules onto the event_server Stream and modify the application's behavior transparently to the other components. There are a bunch of features that aren't implemented in this prototype that you'd probably want to do for a "real" application. Some of the more interesting things to add would be: 0. Complete "full-duplex" support, i.e., Peers could play the role of Suppliers and Consumers simultaneously. 1. Support for "commands", which would change the behavior of the Event_Server based on messages it got from Suppliers (or Consumers). 3. Support for "pull" operations, as well as "push" operations. This would basically involve adding a "MIB Module" to get/set the "values" associated with "names" passed in by Peers. This could probably replace the Event_Analysis Module. 4. Filtering and correlation (this should probably be done via a separate Module that handles filtering and correlation). 5. More flexible concurrency model(s), e.g., "Active Object per-Consumer". This would enable the Event Server process to handle flow control more gracefully than it does not (it currently "hangs," which isn't desirable).