No, the test is single-threaded. Multi-threaded version is easy to create, but I don't see much sense in it now, until a single-threaded test starts to pass stably. Also, multi-threaded test crashes would be more difficult to analyze.
Yes, I have reduced the test, looks like the problem is not in mass class loading, but in org.mmtk.plan.Plan class itself. Both RI and DRLVM terminate without catching exception that occurs on this class's initialization. Really strange. See http://issues.apache.org/jira/browse/HARMONY-2650 Vasily -----Original Message----- From: news [mailto:[EMAIL PROTECTED] On Behalf Of Egor Pasko Sent: Tuesday, December 12, 2006 5:11 PM To: [email protected] Subject: Re: [drlvm][test] Classloading stress test available On the 0x23C day of Apache Harmony Vasily M. Zakharov wrote: > Hi, all, > > I've just uploaded a classloading stress test to > http://issues.apache.org/jira/browse/HARMONY-2615. > It locates all the classes available in bootclasspath and tries to load > each of them with a simple Class.forName() call. does it run in multiple threads (for better stressing:)? > This simple test already helped to find a number of bugs in DRLVM and > may potentially find some more, so I recommend including it to some test > suite that is run regularly. > > By the way, the test fails on the current build, and the fail looks very > strange - at some point, an uncaught exception emerges from inside the > catch(Throwable) block. interesting! could you, please, reduce the test? -- Egor Pasko
