bug in Integer.decode ?

[Enno Rehling]

I couldn't resist. There's a bug in kaffe's Integer.java, just as I
expected: when the Integer is not prefixed by 0x, O or #, it's not correct
to do nm.substring(1), just nm is okay.

What's the proper procedre for reporting bugs anyway? I'm pretty new to
kaffe, but I will probably be working with a lot it in the near future.

Also, are there benchmarks for a VM that compare to C++ programs? I know
Caffeinmark, but that only gives me performance relative to other VMs. I'm
interested in performance relative to C++.

Enno.