Code Review for WeakReference leak in the Logging API (6942989)
mandy.chung at oracle.com
Mon Jun 14 10:11:10 PDT 2010
Daniel D. Daugherty wrote:
> On 6/11/2010 2:09 PM, Mandy Chung wrote:
>> Daniel D. Daugherty wrote:
>>> The theory is that there shouldn't be too many Logger objects in a
>>> normal system and
>>> once they have been added, then this fix doesn't come into play.
>>> I would be surprised if a real system had more than 100 Logger
>> FYI. AWT creates a number of loggers (see
>> Running a simple Framer application (awt helloworld) with JDK 7 b70
>> creates 79 loggers on solaris-i586 and 34 loggers on windows-i586.
>> SwingSet2 creates a total of 85 loggers including a few non-awt ones
>> on solaris-i586 and 35 on windows-i586).
> Thanks for the numbers. I still don't consider these numbers of
> Loggers to be a lot.
These numbers are not a lot. Just to show as an example that there
might be applications out there doing similar thing (one Logger per
class) that you can take into consideration.
> However, the point will be possible moot
> depending on what happens with my investigation of ReferenceQueues
> and Jeremy Manson's fix for this same problem.
> Stay tuned!
> Mandy, did you also review the code?
No, I didn't.
More information about the hotspot-runtime-dev