> Oh yes, I could use this third approach, but in that case I need to > write the "retrieval routines" two times: in HTTP and in Net-SNMP > agent.
If you can afford the overhead, and the data is small enough that RS232's limited bandwidth won't be a bottleneck, why don't you write a simple common library, and just let each get data as they need it? Then if testing shows that you really need to minimize calls through that port, change its implementation to use one cache with access via IPC/RPC/whatever you feel like, and the interface to both your http daemon and your snmp agent won't need to change. If you need to provide access to other components down the road it's simpler, too, and without having dependancies between httpd and snmpd you won't need to have one running for the other to work. Unless you have strict latency requirements, worrying about that beforehand is probably a red herring. If latency really is a big deal, either push from or continuously poll your device. -Dan ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Net-snmp-coders mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
