Re: Potential Software Failure

The code looks ok, it should work this way, yet without an
example I can't verify. Check your back up drives ... ;-)

Cheers
	Markus

David Dixon-Peugh wrote:

> I don't have the file which has the problem here
> (forgot to check it into CVS yesterday.)
> 
> The code that it generates is exactly like I said
> before.  (Written so as not to be confusing about
> the catch block.  Positions made up.)
> 
> 
> 0   ACONST_NULL
> 1   ICONST( 0 )
> 2   INVOKEVIRTUAL ( runTest(I)I )
> 3   ICONST( 0 )
> 4   IRETURN
> 5   ICONST( 0 )
> 6   IRETURN
> 
> CATCH: From 0
>          To 4
>     Handler 5
>         Exc java.lang.NullPointerException
> 
> The catch block is on the InstructionList, so
> that isn't the problem.
> 
> The code is supposed to do the following:
>   Call "runTest(I)I" on NULL.
>   Catch the NullPointerException.
>   Return 0 from the handler.
> 
> It definitely throws the NPE, and it never
> gets caught.  (Once again, looking at the Method
> that was generated, it didn't have any catch tables.)
> 
> DDP
> 
> 
> --- Markus Dahm <[email protected]> wrote:
> 
>>David Dixon-Peugh wrote:
>>
>>
>>>I have found what I believe to be a bug in BCEL,
>>>and would like to see if anyone else has seen
>>>this in the wild.
>>>
>>>Basically, I'm generating code that looks like
>>>
>>this:
>>
>>>  ACONST_NULL
>>>  ICONST( 0 )
>>>  INVOKEVIRTUAL ( runTest(I)I )
>>>  ICONST( 0 )
>>>  IRETURN
>>>[catch NullPointerException]
>>>  ICONST( 0 )
>>>  IRETURN
>>>
>>>The idea is to force a NullPointerException to be
>>>raised, and then catch it with a Catch clause.
>>>
>>>This is being done in a MethodGen object, using
>>>"addExceptionHandler(...)" to create the 
>>>CodeExceptionGen inside the MethodGen.  (Adding a
>>>CodeExceptionGen is a good idea.  We should be