Path: csiph.com!x330-a1.tempe.blueboxinc.net!usenet.pasdenom.info!gegeweb.org!eternal-september.org!feeder.eternal-september.org!.POSTED!not-for-mail
From: markspace <-@.>
Newsgroups: comp.lang.java.programmer
Subject: Re: Making one or more threads wait for another to produce a value or fail
Date: Thu, 02 Jun 2011 14:10:28 -0700
Organization: A noiseless patient Spider
Lines: 64
Message-ID: <is8u8e$5jn$1@dont-email.me>
References: <alpine.DEB.2.00.1105311443530.28134@urchin.earth.li>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Thu, 2 Jun 2011 21:10:39 +0000 (UTC)
Injection-Info: mx04.eternal-september.org; posting-host="fblHIPffyJ67C5r/ar9cfA"; logging-data="5751"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX186N8eVtsAaJitgRjMdpDtu4LZhRVD+UBc="
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.17) Gecko/20110414 Thunderbird/3.1.10
In-Reply-To: <alpine.DEB.2.00.1105311443530.28134@urchin.earth.li>
Cancel-Lock: sha1:TLelXwi31RWwT4gBw3rR7UCulI4=
Xref: x330-a1.tempe.blueboxinc.net comp.lang.java.programmer:4902

On 5/31/2011 7:00 AM, Tom Anderson wrote:
> The scenario:
>
> Penelope is a widow, or at least her husband isn't around any more
> (she's not sure which; long story). There are 108 suitors who would like
> to marry her. She hasn't decided which one she'll marry. So, the 108
> suitors are sitting about waiting for her to decide. It's possible that
> instead of deciding to marry one of them, she'll deliver some other,
> exceptional, verdict (eg "turns out my husband is still alive, and will
> now murder you all").


Just popping in quickly.  First, everyone's all about CountDownLatch. 
Does this really work?  It requires a fixed number of threads supplied 
on creation to "fire," iirc.

Second what about other java.util.concurrent classes?  Both locks (esp 
ReadWriteLock) and AbstractQueuedSynchronizer look promising.

<http://download.oracle.com/javase/6/docs/api/java/util/concurrent/locks/AbstractQueuedSynchronizer.html>

Especially the second example for AQS, which is "a latch class that is 
like a CountDownLatch except that it only requires a single signal to 
fire. Because a latch is non-exclusive, it uses the shared acquire and 
release methods."


Lastly, I must be dense, I have a simple implementation but I think it's 
wrong.  What is wrong with it?  I'm sure I've missed something in the 
original specification, but I'm drawing a blank.  How is the exceptional 
verdict delivered?  There's also no way to set a verdict for one thread 
vs setting a return value for all threads.  I think your initial 
specification of using just "await()" and "deliver()" might be a bit too 
simple for the problem.

This is just a simple class that sends a single message (the verdict) to 
all threads (i.e., it broadcasts the message).  Note this is untested.


public class BroadcastSynchronizer<V, E extends Throwable> {

     private volatile V verdict;
     private final Object lock = new Object();

     public V await() throws E, InterruptedException {
         while( verdict == null ) {
             synchronized( lock ) {
                 if( verdict == null )
                     lock.wait();
             }
         }
         return verdict;
     }

     public void deliver( V verdict ) {
         this.verdict = verdict;
         synchronized( lock ) {
            lock.notifyAll();
         }
     }
}