libgcj/215

[Tom Tromey]

The following reply was made to PR libgcj/215; it has been noted by GNATS.

From: Tom Tromey <tromey@cygnus.com>
To: Bryce McKinlay <bryce@albatross.co.nz>
Cc: green@cygnus.com, java-gnats@sourceware.cygnus.com
Subject: Re: libgcj/215
Date: Tue, 25 Apr 2000 15:13:16 -0700 (PDT)

 Bryce>  Basically, loading native code from interpreted code doesn't
 Bryce>  yet work.  (In this case loading a native method
 Bryce>  implementation, but loading compiled java code using forName()
 Bryce>  from interpreted code doesn't work yet, either).
 
 I've solved this problem when the native method is implemented using
 JNI (patch pending).  But we haven't even considered it for CNI code.
 I guess there's no reason to prohibit this case, though it does seem
 unusual.  One idea would be to add a CNI function that could be used
 to register these functions (from JNI_OnLoad).  Another idea would be
 to hack jni.cc to generate the mangling for such methods and look for
 them before looking for a CNI method.  Ouch.
 
 Tom