Re: Patches for faster interface dispatch, type checking

[Godmar Back <gback at cs dot utah dot edu>]

> 
> Godmar Back <gback@cs.utah.edu> writes:
> 
> > From a quick look at it though, I don't think it's 100% correct.
> > I don't see any checks for NoSuchMethodError and I also don't see
> > how you handle the java.lang.Object case.  
> 
> Well, it does look like we need to enhance the verifier (both
> in the compiler and the interpreter) to handle interfaces as discussed
> earlier.  While this is only needed to catch bogus bytecode, and isn't
> top-critical, it does need to be done.
> 

Not throwing NoSuchMethodErrors is not top-critical if you assume a
consistent compilation environment.

Not being able to handle java.lang.Object is maybe not top-critical,
but it's important in that it currently prevents gcj from executing valid
bytecode other JVMs do execute.  (I.e., bytecode you may find in a binary
distribution of a software package.)

	- Godmar