RFR 8028623: SA: hash codes in SymbolTable mismatching java_lang_String::hash_code for extended characters.

[Kevin Walls]

Thanks - just need that second reviewer then...


On 13/01/14 12:38, Staffan Larsen wrote:
> Looks good! Thanks for taking the time to re-write the test in Java.
>
> Thanks,
> /Staffan
>
> On 8 jan 2014, at 15:59, Kevin Walls <kevin.walls at oracle.com> wrote:
>
>> Hi Staffan -
>>
>> http://cr.openjdk.java.net/~kevinw/8028623/webrev.01/
>>
>> Yes it's better now, getting pid and launching a tool etc have been previous reasons to use a script, but with this new help it's not too bad!...
>>
>> Thanks,
>> Kevin
>>
>>
>> On 07/01/14 09:43, Staffan Larsen wrote:
>>> Kevin,
>>>
>>> The fix looks good.
>>>
>>> For tests, we are trying to avoid adding new shell-script based tests since they too often cause problems. Would it be possible to rewrite the test in pure Java code? There are some helper routines in test/testlibrary/com/oracle/java/testlibrary/ that could be useful.
>>>
>>> /Staffan
>>>
>>> On 3 jan 2014, at 18:42, Kevin Walls <kevin.walls at oracle.com> wrote:
>>>
>>>> Hi,
>>>>
>>>> This problem means you can't use the SA if the target app contains a symbol which uses a non-ascii character.  The SA tool will fail with an error, the JVM itself and the SA having calculated different hashes for such Strings.
>>>>
>>>> bug:
>>>> https://bugs.openjdk.java.net/browse/JDK-8028623
>>>>
>>>> webrev:
>>>> http://cr.openjdk.java.net/~kevinw/8028623/webrev.00/
>>>>
>>>> Thanks
>>>> Kevin