Does anybody have any numbers on the performance impact of using the "M" conversion character? We recently had a problem in an application where the overall application performance was affected by a factor of 500% by the use of a stack trace internally by one component. I would expect that "M" would have a similar impact. Creating exceptions is relatively light but printStackTrace is extremely expensive.
I would be extremely hesitant to use this approach. If the method name is absolutely required, it may be better to design a log4j wrapper that mandates the use of a methodName argument. -----Original Message----- From: RUTHERFURD, Michael [mailto:[EMAIL PROTECTED] Sent: Wednesday, 4 June 2003 11:18 AM To: Log4J Users List Subject: RE: Conversion Character M I've had this problem running WebSphere 3.02 on an AS400. In my case it was caused by an non-standard stack trace. I customised the parsing and the problem went away. Michael Rutherfurd > i am trying to use the conversion character "M" for outputting the method > name where the logging request was issued. > unfortunately all log4j wirtes to the logfile is ?... the conversion > configuration is the following: > > value="%d{dd.MM.yyyy HH:mm:ss} %r [%t] *%-5p %-17c{2} [%M] - %m%n" > > the log-statements are all of the following: > > 02.06.2003 10:45:52 5508343 [HttpListener-1:Thread-3] *DEBUG > persistence.AbstractDBMapper [?] - Method execute called... > > any ideas? > > kind regards, > sandro --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]