On Tue, 2006-02-21 at 20:51 +0000, robert burrell donkin wrote: > On Tue, 2006-02-21 at 00:42 +0100, Remy Maucherat wrote: > > On 2/21/06, robert burrell donkin <[EMAIL PROTECTED]> wrote: > > > i've been running some performance tests: the aim being to check how > > > much slower JCL 1.1 runs that the 1.0.x releases. the figures for log4j > > > are interesting. the raw data is below (the percentages are normalised > > > to 1.0.4 benchmark times). > > > > > > 1. there is a small performance degradation between the 1.0.3, 1.0.4 and > > > 1.1RC5 but not significant (a few percentiles). > > > 2. 1.0.2 is a *lot* faster for log4j that the later releases > > > 3. isDebugEnabled is a *lot* faster than logging > > > 4. getLog is *slow* > > > > > > i have some ideas for improving performance against log4j (but testing > > > them will have to wait till tomorrow). > > > > As long as isDebugEnabled is fast, I'm happy. In most cases, debug > > level will mean logging a ton of things, so will be slow (of course, > > people have complained that TC was too slow with debug logging). > > i probably should have mentioned that the tests were for debugging to a > disabled category/logger (whichever nomenclature you prefer): something > smells bad :-/
and it was my configuration: log4j is *very* slow which it's not explicitly configured. with a log4j file, the figures are pretty much what i'd expect. that's a relief :) i'm not sure why JCL 1.0.2 is quicker a configuration file. it does use category (rather than logger)... - robert --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
