thanks to you both for your investigations i'm not sure that examples 5 and 6 support the conclusions as strongly as they might. log4jlogger contains a symbolic link to class distributed as part of log4j. it must therefore be placed in a classloader that can load log4j. therefore the NoClassDefFoundError is required. probably better to use the api jar in the parent and main jar in the child.
some of the rest of ceki's are interesting :) the behaviour in brian's example 10 seems a little unexpected... i've mainly been going through the code by eye so far. i'd really like to get some proper unit test to try to nail down whether some of ceki's points are down to simple implementation flaws. i'd like to address brian's comments about packaging separately (since it relates to JCL2, in my mind if nowhere else) - robert On Thu, 2005-03-03 at 18:05, Ceki G�lc� wrote: > Brian, > > Thank you for your analysis which is very much appreciated. I think it > deserves further investigation. Unfortunately, I can't spare the time to > study it right away, but promise to do so as soon as my schedule clears up. > > On 2005-03-03 5:34:50, Brian Stansberry wrote: > > > What I've found is documented at > > http://xnet.wanconcepts.com/jcl/furtherAnalysis.html. > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
