Re: JNI patches

[mgd at swarm dot org (Marcus G. Daniels)]

>>>>> "PB" == Per Bothner <per@bothner.com> writes:

PB> I'm not seeing how a valid JNI program can not work given this
PB> implementation.  I'm assuming the conservative GC will find both
PB> local and global references assuming it normally scans C static
PB> and local variables. 

I'm just saying that the existing implementation (which does not have
NOPs for Delete{Local,Global}Ref), doesn't work.  While the JNI
spec warns that:

  Note that local references cannot be faithfully implemented by
  conservatively scanning the native stack. The native code may store
  local references into global or heap data structures.

..but if the conservative GC takes the heap into the account, it seems
that should work.  Anyway, I just want it not to crash.  ;)