On 31/08/2013 3:04 AM, Jason Mehrens wrote:
The coarse grain locking in FileHandler, MemoryHandler, SocketHandler, and
StreamHandler publish could/should be relaxed so we are not calling isLoggable
while holding the lock.
I'm not sure I'd agree with that. If you moved isLoggable outside the
sync region you could get inconsistent logging results if the log level
is changed after the check - eg thread A is logging while the level is
'high' and has checked isLoggable and it returned true and it gets
preempted. Thread B changes the log level and adds a record to the log
indicating that. Thread A then continues and you appear to get a "wrong"
log entry.
Anyway definitely a future RFE not something to be modified for this issue.
Cheers,
David
Jason
Date: Fri, 30 Aug 2013 13:50:56 +0200
From: daniel.fu...@oracle.com
To: david.hol...@oracle.com
Subject: Re: RFR: JDK-6823527: java.util.logging.Handler has thread safety
issues
CC: core-libs-dev@openjdk.java.net
Hi,
Please find below an updated patch for solution (c)
<http://cr.openjdk.java.net/~dfuchs/webrev_6823527/webrev.02/>
best regards,
-- daniel
