Package de.aristaflow.adept2.util.threading.executor

Class ObjectOneWorkRunnable<O,R extends AbstractObjectOneWorkExecutor.RuntimeData>

java.lang.Object

de.aristaflow.adept2.util.threading.executor.AbstractObjectRunnable<O,O,R>

de.aristaflow.adept2.util.threading.executor.ObjectOneWorkRunnable<O,R>

Type Parameters:

O - The type of object for which to run.

R - The runtime data for managed objects. This is a tuple of pending work before and after a flush.

All Implemented Interfaces:

ObjectRunnable<O,O>, Runnable

public class ObjectOneWorkRunnable<O,R extends AbstractObjectOneWorkExecutor.RuntimeData>
extends AbstractObjectRunnable<O,O,R>

A runnable for having exactly one runnable per object. It removes the object from all work before running and thus allows to add a new runnable for this object.

Field Summary

Fields

Modifier and Type	Field	Description
protected final AbstractObjectOneWorkExecutor<O,R>	executor	The executor to which to add this ObjectRunnable in case there is more work for the object.
protected final AtomicInteger	flushPending	The amount of pending flushs.

Fields inherited from class de.aristaflow.adept2.util.threading.executor.AbstractObjectRunnable

logger, objects, terminate

Constructor Summary

Constructors

Modifier	Constructor	Description
protected	ObjectOneWorkRunnable(O object, AbstractObjectOneWorkExecutor<O,R> executor, AtomicInteger terminate, AtomicInteger flushPending, Logger logger)	Creates a new runnable that executes one runnable for a specific object.

Method Summary

Modifier and Type	Method	Description
protected boolean	runForObject()	Runs this runnable actually.

Methods inherited from class de.aristaflow.adept2.util.threading.executor.AbstractObjectRunnable

awaitTermination, getIndexObject, getRegisteredObject, run

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

executor

protected final AbstractObjectOneWorkExecutor<O,R extends AbstractObjectOneWorkExecutor.RuntimeData> executor

The executor to which to add this ObjectRunnable in case there is more work for the object. Re-adding instead of executing the work prevents starvation of work for specific objects.

flushPending

protected final AtomicInteger flushPending

The amount of pending flushs. If a flush is pending, work for the same object has to be executed by the same runnable (and thread) instead of re-scheduling it.
Normally, re-schedlung prevents starvation of work. When a flush is pending, the current work will be executed before the flush so no starvation will take place since the amount of work (before the flush) is restricted.
Access has to be synchronised via AbstractObjectRunnable.objects.

Constructor Details

ObjectOneWorkRunnable

protected ObjectOneWorkRunnable(O object,
 AbstractObjectOneWorkExecutor<O,R> executor,
 AtomicInteger terminate,
 AtomicInteger flushPending,
 Logger logger)

Creates a new runnable that executes one runnable for a specific object.

Parameters:

object - The object of this runnable.

executor - The executor to which to add this ObjectRunnable in case there is more work for the object.

terminate - The amount of termination requests. This allows for synchronising across different objects.

flushPending - The amount of pending flushs. If a flush is pending, work for the same object has to be executed by the same runnable (and thread) instead of re-scheduling it.

logger - The logger which to use for runtime exceptions stemming from a executing a runnable. Such an exception will only be logged.

Method Details

runForObject

protected boolean runForObject()

Description copied from class: AbstractObjectRunnable

Runs this runnable actually. Activation and termination signalling are already handled, implementations only need to react to termination requests.

Specified by:

runForObject in class AbstractObjectRunnable<O,O,R extends AbstractObjectOneWorkExecutor.RuntimeData>

Returns:

Whether the work for the object has been terminated. Otherwise there is work left and the termination is not signalled. With a termination request, return true, only return false if you want to continue to work without interruption, e.g. to change the executing thread.