I tested the same snippet with log4j and it runs fast with javaw on windows xp (700ms). The versions were log4j 1.2.15; org.slf4j.api 1.15.11 and org.slf4j.log4j12 1.4.2 The slow configuration was: org.slf4j.api 1.15.11 and org.slf4j.log4j12 1.4.2 ch.qos.logback.classic, ch.qos.logback.core, ch.qos.logback.slf4j all in version 0.9.19. Peter Am 21.12.2011 um 17:53 schrieb ceki:
Thank you for the output of visualvm profiler. It looks like 90.4% of the time is spent in java.io.PrintStream.write(int) called by ch.qos.logback.core.joran.spi.ConsoleTarget. There is also some time spent in java.io.IOException.<init>(String), i.e the constructor of IOException.
In principle, the performance of log4j ConsoleAppender should be very similar, although its execution path might be slightly shorter.
Which version of log4j are you using?
-- Ceki http://twitter.com/#!/ceki
On 21.12.2011 16:04, Peter Kullmann wrote:
I don't know what to do except perhaps warn against the usage of the console appender when running without a console on windows xp.
I'm attaching the output of the visualvm profiler so that you can see where all the time is being spent.
By the way, I think log4j doesn't have this problem. We were using log4j for years in much the same setups.
Peter
Am 21.12.2011 um 11:03 schrieb ceki:
On 21.12.2011 10:55, Peter Kullmann wrote:
Hi Ceki,
The problem manifests itself under windows xp. Windows 7 is much better.
I don't see anything actionable to work with here. Ideas?
Peter
_______________________________________________ Logback-user mailing list Logback-user@qos.ch http://mailman.qos.ch/mailman/listinfo/logback-user