> On May 25, 2017, at 12:24 AM, David Holmes <david.hol...@oracle.com> wrote: > > Correction ... > > On 25/05/2017 5:00 PM, David Holmes wrote: >> Hi Mandy >> On 25/05/2017 8:10 AM, Mandy Chung wrote: >>> http://cr.openjdk.java.net/~mchung/jdk9/webrevs/8181033/webrev.00/ >>> >>> This fixes the launcher to print the exception being thrown rather than >>> the confusing message "Error: A JNI error has occurred, please check >>> your installation and try again”. A few attempts to improve this to >>> catch ClassNotFoundException and LinkageError. This catches other cases. >> Based on the exception stack that Volker reported this looks fine - it >> catches the unexpected exceptions encountered during getMethod() and reports >> them. >> One thing that confuses me though, is that Volker reported the problem as >> occurring due to verification - and he showed some VM log output to support >> that. But verification happens when the main class is loaded! The exception >> trace shows code that executes well after loading time. ?? > > Never mind. The Class.forName call requests the class not be initialized, > which means it will not be linked. Verification is actually part of linking. > The class has to be initialized before getMethod can be invoked, and hence it > must first be linked and thus verified.
Yes exactly. FWIW [1][2] are references to linking and initialization. Mandy [1] https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-5.html#jvms-5.4 [2] https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-5.html#jvms-5.5
