Bug#6546113 StringCharBuffer fix

[Alan Bateman]

Roman Kennke wrote:
> Hi there
>
>   
>>> This fixes a problem with sliced StringCharBuffers. See:
>>>
>>> http://bugs.sun.com/view_bug.do?bug_id=6546113
>>>
>>> Feel free to integrate this in the repository.
>>>       
>
> Ping, any opinions on this?
>
> /Roman
>   
I think Iris has been meaning to get back to you on this for a while 
(she has been busy). It's a bug that she & I spoke about a few months 
ago and she had planned to fix. All the better that you are now taking 
time to address it. If you look at 4997655 you will see that slicing a 
CharBuffer that wraps a char sequence has been problematic since day 
one. The changes for that bug in jdk6 fixed some problems but also lead 
to this problem. Unfortunately, the bug wasn't caught by existing tests 
so one thing that would be good to do as part of 6546113 is to add or 
improve the regression tests for char buffers that wrap a char sequence. 
Are you interested in doing that as part of the fix to this bug? Anyway, 
your changes look like a reasonable fix for the get methods. Did you 
check the toString, subSequence, and duplicate methods in case they have 
the same problem?  One small concern is that the offset field was 
originally the offset into the backing array for heap buffers. Now it is 
used as the offset into the char sequence in the wrapped case. We should 
probably comment that for the sake of future maintainers or else add a 
new strOffset field to StringCharBuffer to make it easier to understand.

-Alan.