Re: LocationInfo performance, StackTraceElement

2004-09-08 Thread Ceki Gülcü
Hello Martin, At 03:40 AM 5/18/2004, Martin Schulz wrote: For example: jakarta-log4j-1.2.8/src/java/org/apache/log4j/performance/xml/logging13.xml 10 gives me 82 us vs 183 us before. (AMD Athlon 1600 at 1.3GHz, JDK 1.4.2_04 on Win XP) I supposed that's the pure speedup ignoring any real effo

Re: LocationInfo performance, StackTraceElement

2004-05-17 Thread Martin Schulz
Ceki, thanks for the encouragement ;-) Ceki Gülcü wrote: Hello Martin, Thanks for the code. What kind of improvement are we talking about? Do you have figures by any chance? We should use the Throwable.getStackTrace() method when running with JDK 1.4 and later. Doing this while keeping *compile*

Re: LocationInfo performance, StackTraceElement

2004-05-17 Thread Ceki Gülcü
Hello Martin, Thanks for the code. What kind of improvement are we talking about? Do you have figures by any chance? We should use the Throwable.getStackTrace() method when running with JDK 1.4 and later. Doing this while keeping *compile* time compatibility with JDK 1.2 is a little tricky but

RE: LocationInfo performance, StackTraceElement

2004-05-16 Thread Paul Smith
> I have been profiling log4j a little bit and found that > LocationInfo (if > used in the pattern) is a particular > bottleneck. I find it amazing that nobody seems to have > suggested yet > to use Throwable.getStackTrace() > instead of parsing the output of the printout. (as of JDK 1.4). >