io.reactivex.schedulers

Class Schedulers

java.lang.Object

io.reactivex.schedulers.Schedulers

public final class Schedulers
extends Object

Static factory methods for returning standard Scheduler instances.

The initial and runtime values of the various scheduler types can be overridden via the RxJavaPlugins.setInit(scheduler name)SchedulerHandler() and RxJavaPlugins.set(scheduler name)SchedulerHandler() respectively.

Supported system properties (System.getProperty()):

• rx2.io-keep-alive-time (long): sets the keep-alive time of the io() Scheduler workers, default is IoScheduler.KEEP_ALIVE_TIME_DEFAULT
• rx2.io-priority (int): sets the thread priority of the io() Scheduler, default is Thread.NORM_PRIORITY
• rx2.io-scheduled-release (boolean): true sets the worker release mode of the io() Scheduler to scheduled, default is false for eager mode.
• rx2.computation-threads (int): sets the number of threads in the computation() Scheduler, default is the number of available CPUs
• rx2.computation-priority (int): sets the thread priority of the computation() Scheduler, default is Thread.NORM_PRIORITY
• rx2.newthread-priority (int): sets the thread priority of the newThread() Scheduler, default is Thread.NORM_PRIORITY
• rx2.single-priority (int): sets the thread priority of the single() Scheduler, default is Thread.NORM_PRIORITY
• rx2.purge-enabled (boolean): enables periodic purging of all Scheduler's backing thread pools, default is false
• rx2.purge-period-seconds (int): specifies the periodic purge interval of all Scheduler's backing thread pools, default is 1 second
• rx2.scheduler.use-nanotime (boolean): true instructs Scheduler to use System.nanoTime() for Scheduler.now(TimeUnit), instead of default System.currentTimeMillis() (false)

Method Summary

Modifier and Type	Method and Description
static Scheduler	computation() Returns a default, shared Scheduler instance intended for computational work.
static Scheduler	from(Executor executor) Wraps an Executor into a new Scheduler instance and delegates schedule() calls to it.
static Scheduler	from(Executor executor, boolean interruptibleWorker) Wraps an Executor into a new Scheduler instance and delegates schedule() calls to it.
static Scheduler	io() Returns a default, shared Scheduler instance intended for IO-bound work.
static Scheduler	newThread() Returns a default, shared Scheduler instance that creates a new Thread for each unit of work.
static void	shutdown() Shuts down the standard Schedulers.
static Scheduler	single() Returns a default, shared, single-thread-backed Scheduler instance for work requiring strongly-sequential execution on the same background thread.
static void	start() Starts the standard Schedulers.
static Scheduler	trampoline() Returns a default, shared Scheduler instance whose Scheduler.Worker instances queue work and execute them in a FIFO manner on one of the participating threads.

Methods inherited from class java.lang.Object

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

Method Detail

computation

@NonNull
public static Scheduler computation()

Returns a default, shared Scheduler instance intended for computational work.

This can be used for event-loops, processing callbacks and other computational work.

It is not recommended to perform blocking, IO-bound work on this scheduler. Use io() instead.

The default instance has a backing pool of single-threaded ScheduledExecutorService instances equal to the number of available processors (Runtime.availableProcessors()) to the Java VM.

Unhandled errors will be delivered to the scheduler Thread's Thread.UncaughtExceptionHandler.

This type of scheduler is less sensitive to leaking Scheduler.Worker instances, although not disposing a worker that has timed/delayed tasks not cancelled by other means may leak resources and/or execute those tasks "unexpectedly".

If the RxJavaPlugins.setFailOnNonBlockingScheduler(boolean) is set to true, attempting to execute operators that block while running on this scheduler will throw an IllegalStateException.

You can control certain properties of this standard scheduler via system properties that have to be set before the Schedulers class is referenced in your code.

Supported system properties (System.getProperty()):

• rx2.computation-threads (int): sets the number of threads in the computation() Scheduler, default is the number of available CPUs
• rx2.computation-priority (int): sets the thread priority of the computation() Scheduler, default is Thread.NORM_PRIORITY
• rx2.io-scheduled-release (boolean): true sets the worker release mode of the #io() Scheduler to scheduled, default is false for eager mode.

The default value of this scheduler can be overridden at initialization time via the RxJavaPlugins.setInitComputationSchedulerHandler(io.reactivex.functions.Function) plugin method. Note that due to possible initialization cycles, using any of the other scheduler-returning methods will result in a NullPointerException. Once the Schedulers class has been initialized, you can override the returned Scheduler instance via the RxJavaPlugins.setComputationSchedulerHandler(io.reactivex.functions.Function) method.

It is possible to create a fresh instance of this scheduler with a custom ThreadFactory, via the RxJavaPlugins.createComputationScheduler(ThreadFactory) method. Note that such custom instances require a manual call to Scheduler.shutdown() to allow the JVM to exit or the (J2EE) container to unload properly.

Operators on the base reactive classes that use this scheduler are marked with the @SchedulerSupport(COMPUTATION) annotation.

When the Scheduler.Worker is disposed, the underlying worker can be released to the cached worker pool in two modes:

• In eager mode (default), the underlying worker is returned immediately to the cached worker pool and can be reused much quicker by operators. The drawback is that if the currently running task doesn't respond to interruption in time or at all, this may lead to delays or deadlock with the reuse use of the underlying worker.
• In scheduled mode (enabled via the system parameter rx2.io-scheduled-release set to true), the underlying worker is returned to the cached worker pool only after the currently running task has finished. This can help prevent premature reuse of the underlying worker and likely won't lead to delays or deadlock with such reuses. The drawback is that the delay in release may lead to an excess amount of underlying workers being created.

Returns:

a Scheduler meant for computation-bound work

io

@NonNull
public static Scheduler io()

Returns a default, shared Scheduler instance intended for IO-bound work.

This can be used for asynchronously performing blocking IO.

The implementation is backed by a pool of single-threaded ScheduledExecutorService instances that will try to reuse previously started instances used by the worker returned by Scheduler.createWorker() but otherwise will start a new backing ScheduledExecutorService instance. Note that this scheduler may create an unbounded number of worker threads that can result in system slowdowns or OutOfMemoryError. Therefore, for casual uses or when implementing an operator, the Worker instances must be disposed via Disposable.dispose().

It is not recommended to perform computational work on this scheduler. Use computation() instead.

Unhandled errors will be delivered to the scheduler Thread's Thread.UncaughtExceptionHandler.

You can control certain properties of this standard scheduler via system properties that have to be set before the Schedulers class is referenced in your code.

Supported system properties (System.getProperty()):

• rx2.io-keep-alive-time (long): sets the keep-alive time of the io() Scheduler workers, default is IoScheduler.KEEP_ALIVE_TIME_DEFAULT
• rx2.io-priority (int): sets the thread priority of the io() Scheduler, default is Thread.NORM_PRIORITY

The default value of this scheduler can be overridden at initialization time via the RxJavaPlugins.setInitIoSchedulerHandler(io.reactivex.functions.Function) plugin method. Note that due to possible initialization cycles, using any of the other scheduler-returning methods will result in a NullPointerException. Once the Schedulers class has been initialized, you can override the returned Scheduler instance via the RxJavaPlugins.setIoSchedulerHandler(io.reactivex.functions.Function) method.

It is possible to create a fresh instance of this scheduler with a custom ThreadFactory, via the RxJavaPlugins.createIoScheduler(ThreadFactory) method. Note that such custom instances require a manual call to Scheduler.shutdown() to allow the JVM to exit or the (J2EE) container to unload properly.

Operators on the base reactive classes that use this scheduler are marked with the @SchedulerSupport(IO) annotation.

Returns:

a Scheduler meant for IO-bound work

trampoline

@NonNull
public static Scheduler trampoline()

Returns a default, shared Scheduler instance whose Scheduler.Worker instances queue work and execute them in a FIFO manner on one of the participating threads.

The default implementation's Scheduler.scheduleDirect(Runnable) methods execute the tasks on the current thread without any queueing and the timed overloads use blocking sleep as well.

Note that this scheduler can't be reliably used to return the execution of tasks to the "main" thread. Such behavior requires a blocking-queueing scheduler currently not provided by RxJava itself but may be found in external libraries.

This scheduler can't be overridden via an RxJavaPlugins method.

Returns:

a Scheduler that queues work on the current thread

newThread

@NonNull
public static Scheduler newThread()

Returns a default, shared Scheduler instance that creates a new Thread for each unit of work.

The default implementation of this scheduler creates a new, single-threaded ScheduledExecutorService for each invocation of the Scheduler.scheduleDirect(Runnable) (plus its overloads) and Scheduler.createWorker() methods, thus an unbounded number of worker threads may be created that can result in system slowdowns or OutOfMemoryError. Therefore, for casual uses or when implementing an operator, the Worker instances must be disposed via Disposable.dispose().

Unhandled errors will be delivered to the scheduler Thread's Thread.UncaughtExceptionHandler.

You can control certain properties of this standard scheduler via system properties that have to be set before the Schedulers class is referenced in your code.

Supported system properties (System.getProperty()):

• rx2.newthread-priority (int): sets the thread priority of the newThread() Scheduler, default is Thread.NORM_PRIORITY

The default value of this scheduler can be overridden at initialization time via the RxJavaPlugins.setInitNewThreadSchedulerHandler(io.reactivex.functions.Function) plugin method. Note that due to possible initialization cycles, using any of the other scheduler-returning methods will result in a NullPointerException. Once the Schedulers class has been initialized, you can override the returned Scheduler instance via the RxJavaPlugins.setNewThreadSchedulerHandler(io.reactivex.functions.Function) method.

It is possible to create a fresh instance of this scheduler with a custom ThreadFactory, via the RxJavaPlugins.createNewThreadScheduler(ThreadFactory) method. Note that such custom instances require a manual call to Scheduler.shutdown() to allow the JVM to exit or the (J2EE) container to unload properly.

Operators on the base reactive classes that use this scheduler are marked with the @SchedulerSupport(NEW_TRHEAD) annotation.

Returns:

a Scheduler that creates new threads

single

@NonNull
public static Scheduler single()

Returns a default, shared, single-thread-backed Scheduler instance for work requiring strongly-sequential execution on the same background thread.

Uses:

• event loop
• support Schedulers.from(Executor) and from(ExecutorService) with delayed scheduling
• support benchmarks that pipeline data from some thread to another thread and avoid core-bashing of computation's round-robin nature

Unhandled errors will be delivered to the scheduler Thread's Thread.UncaughtExceptionHandler.

This type of scheduler is less sensitive to leaking Scheduler.Worker instances, although not disposing a worker that has timed/delayed tasks not cancelled by other means may leak resources and/or execute those tasks "unexpectedly".

If the RxJavaPlugins.setFailOnNonBlockingScheduler(boolean) is set to true, attempting to execute operators that block while running on this scheduler will throw an IllegalStateException.

You can control certain properties of this standard scheduler via system properties that have to be set before the Schedulers class is referenced in your code.

Supported system properties (System.getProperty()):

• rx2.single-priority (int): sets the thread priority of the single() Scheduler, default is Thread.NORM_PRIORITY

The default value of this scheduler can be overridden at initialization time via the RxJavaPlugins.setInitSingleSchedulerHandler(io.reactivex.functions.Function) plugin method. Note that due to possible initialization cycles, using any of the other scheduler-returning methods will result in a NullPointerException. Once the Schedulers class has been initialized, you can override the returned Scheduler instance via the RxJavaPlugins.setSingleSchedulerHandler(io.reactivex.functions.Function) method.

It is possible to create a fresh instance of this scheduler with a custom ThreadFactory, via the RxJavaPlugins.createSingleScheduler(ThreadFactory) method. Note that such custom instances require a manual call to Scheduler.shutdown() to allow the JVM to exit or the (J2EE) container to unload properly.

Operators on the base reactive classes that use this scheduler are marked with the @SchedulerSupport(SINGLE) annotation.

Returns:

a Scheduler that shares a single backing thread.

Since:

2.0

from

@NonNull
public static Scheduler from(@NonNull
                                      Executor executor)

Wraps an Executor into a new Scheduler instance and delegates schedule() calls to it.

If the provided executor doesn't support any of the more specific standard Java executor APIs, cancelling tasks scheduled by this scheduler can't be interrupted when they are executing but only prevented from running prior to that. In addition, tasks scheduled with a time delay or periodically will use the single() scheduler for the timed waiting before posting the actual task to the given executor.

Tasks submitted to the Scheduler.Worker of this Scheduler are also not interruptible. Use the from(Executor, boolean) overload to enable task interruption via this wrapper.

If the provided executor supports the standard Java ExecutorService API, cancelling tasks scheduled by this scheduler can be cancelled/interrupted by calling Disposable.dispose(). In addition, tasks scheduled with a time delay or periodically will use the single() scheduler for the timed waiting before posting the actual task to the given executor.

If the provided executor supports the standard Java ScheduledExecutorService API, cancelling tasks scheduled by this scheduler can be cancelled/interrupted by calling Disposable.dispose(). In addition, tasks scheduled with a time delay or periodically will use the provided executor. Note, however, if the provided ScheduledExecutorService instance is not single threaded, tasks scheduled with a time delay close to each other may end up executing in different order than the original schedule() call was issued. This limitation may be lifted in a future patch.

Starting, stopping and restarting this scheduler is not supported (no-op) and the provided executor's lifecycle must be managed externally:

 ExecutorService exec = Executors.newSingleThreadedExecutor();
 try {
     Scheduler scheduler = Schedulers.from(exec);
     Flowable.just(1)
        .subscribeOn(scheduler)
        .map(v -> v + 1)
        .observeOn(scheduler)
        .blockingSubscribe(System.out::println);
 } finally {
     exec.shutdown();
 }
 

This type of scheduler is less sensitive to leaking Scheduler.Worker instances, although not disposing a worker that has timed/delayed tasks not cancelled by other means may leak resources and/or execute those tasks "unexpectedly".

Note that this method returns a new Scheduler instance, even for the same Executor instance.

Parameters:

executor - the executor to wrap

Returns:

the new Scheduler wrapping the Executor

from

@NonNull
public static Scheduler from(@NonNull
                                      Executor executor,
                                      boolean interruptibleWorker)

Wraps an Executor into a new Scheduler instance and delegates schedule() calls to it.

The tasks scheduled by the returned Scheduler and its Scheduler.Worker can be optionally interrupted.

If the provided executor doesn't support any of the more specific standard Java executor APIs, tasks scheduled with a time delay or periodically will use the single() scheduler for the timed waiting before posting the actual task to the given executor.

If the provided executor supports the standard Java ExecutorService API, canceling tasks scheduled by this scheduler can be cancelled/interrupted by calling Disposable.dispose(). In addition, tasks scheduled with a time delay or periodically will use the single() scheduler for the timed waiting before posting the actual task to the given executor.

If the provided executor supports the standard Java ScheduledExecutorService API, canceling tasks scheduled by this scheduler can be cancelled/interrupted by calling Disposable.dispose(). In addition, tasks scheduled with a time delay or periodically will use the provided executor. Note, however, if the provided ScheduledExecutorService instance is not single threaded, tasks scheduled with a time delay close to each other may end up executing in different order than the original schedule() call was issued. This limitation may be lifted in a future patch.

Starting, stopping and restarting this scheduler is not supported (no-op) and the provided executor's lifecycle must be managed externally:

 ExecutorService exec = Executors.newSingleThreadedExecutor();
 try {
     Scheduler scheduler = Schedulers.from(exec, true);
     Flowable.just(1)
        .subscribeOn(scheduler)
        .map(v -> v + 1)
        .observeOn(scheduler)
        .blockingSubscribe(System.out::println);
 } finally {
     exec.shutdown();
 }
 

This type of scheduler is less sensitive to leaking Scheduler.Worker instances, although not disposing a worker that has timed/delayed tasks not cancelled by other means may leak resources and/or execute those tasks "unexpectedly".

Note that this method returns a new Scheduler instance, even for the same Executor instance.

Parameters:

executor - the executor to wrap

interruptibleWorker - if true the tasks submitted to the Scheduler.Worker will be interrupted when the task is disposed.

Returns:

the new Scheduler wrapping the Executor

Since:

2.2.6 - experimental

shutdown

public static void shutdown()

Shuts down the standard Schedulers.

The operation is idempotent and thread-safe.

start

public static void start()

Starts the standard Schedulers.

The operation is idempotent and thread-safe.