Timothy Stack wrote:
I agree with you. You should use JNI_VERSION_1_1 to make kaffe uses Kaffe_JavaVMInitArgs instead of the new mechanism (which is still experimental).Heya,
it seems that the internal jit tests on CVS head can no longer be built. I've tried to write a patch to fix that, but I just managed to make the test compile & fail. Any idea what's wrong? I've attached my patch.
The problem is with JNI_GetDefaultJavaVMInitArgs() in jni-base.c, the JNI_VERSION_1_2 case does not initialize the args from Kaffe_JavaVMInitArgs like the JNI_VERSION_1_1 case does. So, the minheap/maxheap values are zero and any allocations just fail.
BTW, I'll continue writing new JNI 1.2 testcases to be sure the API is functioning.
Cheers, Guilhem Lavaux.
_______________________________________________ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
