2014-09-30 14:50 GMT+04:00 Emmanuel Bourg <ebo...@apache.org>: > Hi Konstantin, > > Thank you very much for the feedback. > >> I have the following concerns: >> >> 1) Someone was testing Tomcat usage of BCEL and found that using this >> caching did not improve performance, but reduced it for our use case. >> It was reported in the following Bugzilla issue: >> >> https://issues.apache.org/bugzilla/show_bug.cgi?id=56940 > > Do you have more details on the size of the jars used to measure the > performance? Did the test let enough time for the JIT to kick in?
It was Mark Thomas who did the testing. Maybe he will add something to the discussion. AFAIK jars from JIRA web application were used for the test. From comment on commit that removed the cache, the difference was 10-15%, for our copy of BCEL that already had other optimizations (code removal) applied: http://svn.apache.org/r1624476 The test code: http://svn.apache.org/viewvc/tomcat/tc8.0.x/tags/TOMCAT_8_0_14/test/org/apache/tomcat/util/bcel/TesterPerformance.java?view=markup >> 2) Configuration of this cache depends on reading System properties such as >> >> final static boolean BCEL_DONT_CACHE = Boolean.getBoolean("bcel.dontCache"); >> >> It is a bit odd to me to configure a library via system properties. >> >> At least there could be a static setter for that flag, or a static >> setter for a cache instance / a factory class. > > I agree, a system property was probably good for Findbugs if it forks > its own VM, but not for a general purpose library. > Best regards, Konstantin Kolinko --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
