On 2 maj 2013, at 16:33, Robert Field <robert.fi...@oracle.com> wrote:
> On 05/02/13 04:08, Staffan Larsen wrote: >> >> On 2 maj 2013, at 12:58, serguei.spit...@oracle.com wrote: >> >>> >>> If I understand correctly, the default methods are generated by the VM and >>> have no presence in class file. >>> Is that right? >> >> Yes, that is my understanding. > > Hmmm, maybe this is what you are saying, but, default methods most certainly > have a presence in class files. The overpass methods, however, are > internally generated by the VM. Yes, I think what I meant is that the default methods are not present in the class that implements the interface, but they are present in the interface class. /Staffan > > -Robert > >> >>> Then the fix looks good. >> >> Thanks! >> >>> A typo - "be included": >>> 631 // and should not b included in the total count >> >> I'll fix before pushing. >> >> /Staffan >> >> >>> >>> Thanks, >>> Serguei >>> >>> >>> On 5/2/13 1:51 AM, Staffan Larsen wrote: >>>> JvmtiClassFileReconstituter does not recognize some methods as generated >>>> default methods when re-creating a class for retransformation. The new >>>> default methods are handled by generating overpass methods inside the JVM, >>>> when reconstituting a class, these methods should be skipped as they >>>> aren't part of the original class. >>>> >>>> webrev: http://cr.openjdk.java.net/~sla/8008453/webrev.00/ >>>> bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8008453 (not >>>> available yet) >>>> >>>> Testing done: java/lang/instrument, com/sun/jdi, nsk.quick-jvmti.testlist >>>> >>>> I have not written a regression test since this was found by running the >>>> existing java/lang/instrument/IsModifiableClassAgent.java test. >>>> >>>> Thanks, >>>> /Staffan >>> >> >
