On Wednesday 16 August 2006 16:13 Anton Luht wrote: > > For example, how you will test that VM is aborted during start-up with > > NoClassDefFoundError (not with IllegalArgumentException or any other > > exception) if it is impossible to load requested class? > > I think it should be done using some VM tests - maybe using JVMTI. > Parsing output sounds too weak for me - anything that prints > "NoClassDefFoundError args[0]" will do. > > This is too complicated though and I agree that parsing error output > will work in most cases.
I've lost the beginning of the discussion so I've considered a limited number of the emails sent today only. If I understood correctly you want to test a correct exception for a child VM process and consider using JVMTI as the testing solution for such exception. I want to point out that JVMTI spec is quite vague in the description of VM startup events sequence, the only guaranteed sequence statement can be found in [1]. The LIVE phase should begin after the main thread has been started. It says nothing about ClassLoad event for main application class, it may be loaded before the main thread is created, therefore before LIVE phase. The Exception event can and should occur only in LIVE phase, that is after VMInit event has been sent. VM can fail and never send the VMInit event if it can't find main application class or some class like the one supplied via -javaagent option before LIVE phase has begun. So I think this testing approach may fail if it relies on unspecified VM behavior. [1] http://java.sun.com/j2se/1.5.0/docs/guide/jvmti/jvmti.html#VMInit -- Gregory Shimansky, Intel Middleware Products Division --------------------------------------------------------------------- Terms of use : http://incubator.apache.org/harmony/mailing.html To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
