RFR 8173845: JShell API: not patch compatible

Robert Field robert.field at oracle.com
Tue Feb 7 04:22:46 UTC 2017


Webrev with memory related changes factored out:

     http://cr.openjdk.java.net/~rfield/8173845v1.webrev/

-Robert

On 02/03/17 07:54, Robert Field wrote:
>
> On 02/03/17 06:23, Jan Lahoda wrote:
>> Overall, seems OK to me. Regarding the removal of final, I wonder: I 
>> would expect that after the JShell is closed, the whole JShell 
>> instance would be released (so that it would be GCed eventually). Who 
>> is keeping the instance(s)?
>
> Over the course of the test runs, hundreds of instances of JShell, the 
> memory use continuously grows, into hundreds of megabytes, eventually 
> causing out of memory.  With these changes it still grows, but not as 
> fast.  I looked but could not find who is holding the instances.  I'm 
> breaking the memory leak part into a separate bug.
>
> -Robert
>
>>
>> Thanks,
>>    Jan
>>
>> On 3.2.2017 08:13, Robert Field wrote:
>>> Please review....
>>>
>>> Bug:
>>>
>>>      https://bugs.openjdk.java.net/browse/JDK-8173845
>>>
>>> Webrev:
>>>
>>>      http://cr.openjdk.java.net/~rfield/8173845v0.webrev/
>>>
>>> Note: the additional test was the straw that broke the camels back -- I
>>> could not get the tests to run without running out of memory, so, for
>>> them to pass I had to add more memory clean-up.
>>>
>>> Thanks,
>>> Robert
>>>
>



More information about the kulla-dev mailing list