Re: Adding a realtime monitoring console

Hi Kelly, > Assuming the app always attempts communication with registered listeners, > and then must timeout and unregister a dead listener...at that point > aren't we just at my preferred solution: socket communication? I don't know much about sockets, so I couldn't say. If the console will connect via ssh, isn't that a stateful TCP connection? Can you keep that connection open with a "keep-alive" or "heartbeat"? Can the ssh server notify the application when the ssh session dies? How expensive is it for the application to discover a dead listener? > A log can be generated from the history table I have implemented. I'm > developing an application framework and performance is my primary concern. > The file I/O overhead concerns me unless I were able to leverage it. Unless you're confident that performance problems will occur, I would encourage you to completely ignore performance, develop the simplest implementation and test, measuring actual performance. > Do you have any links where I might find the pattern described in more > detail? If nothing else, maybe the article contains ideas that can help > kick-start another idea....Thanks...! http://www.dofactory.com/Patterns/PatternObserver.aspx Just google "event listener pattern" or "observer pattern". Also, you may find Extreme Programming to be useful. http://www.extremeprogramming.org/ Eric