On Mon, 10 Nov 2003, Ceki Gülcü wrote: | | Endre, it seems that you use DEBUG and TRACE for development (bug-hunting). | Others tend to use INFO and DEBUG for development (i.e. bug-hunting). | | When do *you* use INFO?
For informational messages like "User logged in", "User so-and-such added customer this-and-that", or "new contract added" or similar "transactional" events - things that would be nice to know if you'd like to investigate very high-level issues, like "when did this contract get added", or "have user such and so ever logged in?" My point is that if you use log4j both for production logging, and for debugging/development, you need INFO for simply "add lines to the log". The next level is "warn", which is an error-condition, as are the last remaining two (error, fatal). This I have talked a little about in earlier postings too. It is all about verbosity, of course. You'd leave info on for production environments, thus it can't be verbose at all, and only include "important events" - simply browsing around on the system should most probably not generate any info-lines at all, or only one per click. Debug is something that you turn on if you'd like to have more contextual information, knowing that this starts to affect performance. Then, on the part that you believe is faulty, you'd turn on tracing. One of the XML parsers of Apahe (I believe) outputs every single token it parses on debug-level (it uses commons logging - debug and trace both go to debug). Such things should definately instead go to trace, as it takes about 2 extra minutes to start up the server (parsing the config) if left on. Endre. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
