Github user dragonsinth commented on a diff in the pull request:

    https://github.com/apache/curator/pull/47#discussion_r18800916
  
    --- Diff: 
curator-recipes/src/main/java/org/apache/curator/framework/recipes/shared/SharedValue.java
 ---
    @@ -129,29 +127,19 @@ public void setValue(byte[] newValue) throws Exception
          * value is updated. i.e. if the value is not successful you can get 
the updated value
          * by calling {@link #getValue()}.
          *
    +     * @deprecated use {@link #trySetValue(VersionedValue, byte[])} for 
stronger atomicity
    +     * guarantees. Even if this object's internal state is up-to-date, the 
caller has no way to
    +     * ensure that they've read the most recently seen value.
    +     *
          * @param newValue the new value to attempt
          * @return true if the change attempt was successful, false if not. If 
the change
          * was not successful, {@link #getValue()} will return the updated 
value
          * @throws Exception ZK errors, interruptions, etc.
          */
    +    @Deprecated
         public boolean trySetValue(byte[] newValue) throws Exception
         {
    -        Preconditions.checkState(state.get() == State.STARTED, "not 
started");
    -
    -        try
    -        {
    -            VersionedValue<byte[]> localCopy = currentValue.get();
    -            
client.setData().withVersion(localCopy.getVersion()).forPath(path, newValue);
    -            currentValue.set(new 
VersionedValue<byte[]>(localCopy.getVersion() + 1, Arrays.copyOf(newValue, 
newValue.length)));
    -            return true;
    -        }
    -        catch ( KeeperException.BadVersionException ignore )
    -        {
    -            // ignore
    -        }
    -
    -        readValue();
    -        return false;
    +        return trySetValue(currentValue.get(), newValue);
         }
    --- End diff --
    
    No, not really.  The old version used a synchronized block.  This version 
uses a compare-and-set loop.  It's only a difference in how concurrency is 
handled, it's not an algorithm change.  For all practical purposes, the CAS 
loop almost never executes more than once.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---

Reply via email to