darwin and java status

[Bryce McKinlay]

Andreas Tobler wrote:

>
>Bryce McKinlay wrote:
> 
>
>>No. You should not need to set CLASSPATH to use gij (unless you have
>>your own classes of course). What happens when you try to debug this in gdb?
>>
>
>which gij: /Volumes/reserved1/gcctmp/bin/gij
>gij --version: gij (GNU libgcj) version 3.1
>
>gdb gij 
>
>--> set args hello
>--> run
>
>(gdb) bt
>#0  _Z20_Jv_IsAssignableFromPN4java4lang5ClassES2_ (target=0x0,
>source=0x15d62c) at ../../../gcc/libjava/java/lang/Class.h:189
>#1  0x000263fc in _Z20_Jv_IsAssignableFromPN4java4lang5ClassES2_
>(target=0x0, source=0x15d62c) at ../../../gcc/libjava/java/lang/natClass.cc:929
>#2  0x000266a0 in _Jv_CheckCast (c=0x0, obj=0x6333b0) at ../../../gcc/libjava/java/lang/natClass.cc:1018
>#3  0x0004051c in
>_ZN4java3net14URLClassLoader12findResourceEPNS_4lang6StringE
>(this=0x6327d0, name=0x633950) at ../../../gcc/libjava/java/net/URLClassLoader.java:188
>


According to this, URLClassLoader.findResource() is calling 
_Jv_CheckCast with a null class argument. Which is odd. Could it have 
been miscompiled?

I guess setting classpath might cause URLClassLoader to take a different 
code path, which is perhaps why the crash didn't occur there?

regards

Bryce.