I agree, this is a serious problem. A fair amount of my JNI code assumes that you can get the superclass's method through the subclass. --John Keiser > -----Original Message----- > From: Bernd Kreimeier [mailto:[EMAIL PROTECTED]] > Sent: Wednesday, August 05, 1998 5:14 AM > To: [EMAIL PROTECTED] > Subject: here we go. > > > [EMAIL PROTECTED] writes: > > committing. > > Good news: compiles, make check result has not changed. > Good news: does not dump on > > env->NewGlobalRef( (jobject)clazz ); > > anymore. Cool. > > Bad news: the "inherit" problem (related to jclass content) > is not gone. See again > http://www.netaccess.ie/~bk/japhar/ Let me emphasize this: Japhar is *wrong* here. In the JDK, methodID's (and probably any inherited field) are available to JNI with the SubClasses, not only with the SuperClass which provides the method. Bad violation of specs. Pretty please, before venturing into optimizations for Class, Object and String, how about making this conformant first? b. P.S.: Thread.stop() still dumps. Ceterum censeo: javac...
