> Great, I've stuck that in the SourceForge code.  Such an 
> error would be consistent with the bug report, though I can't
> easily test it myself.
> 
> Incidentally, shouldn't the maximums be width-1 and height-1? 
> The code appears to count from 0 here.

Gee ... I guess you're right ... thanks for pointing it out.

Cheers,
  - Andreas