android-priority-jobqueue

Development in this repository is stopped. Future development continues on https://github.com/yigit/android-priority-jobqueue

==========================

Android Priority Job Queue (Job Manager)

Priority Job Queue is an implementation of a Job Queue specifically written for Android to easily schedule jobs (tasks) that run in the background, improving UX and application stability.

It is written primarily with [flexibility][10] & [functionality][11] in mind. This is an ongoing project, which we will continue to add stability and performance improvements.

• Why ?
• The Problem
• Our Solution
• Show me the code
• What's happening under the hood?
• Advantages
• Getting Started
• Version History
• Building
• Running Tests
• [wiki][9]
• Dependencies
• License

Why ?

The Problem

Almost every application does work in a background thread. These "background tasks" are expected to keep the application responsive and robust, especially during unfavorable situations (e.g. limited network connectivity). In Android applications, there are several ways to implement background work:

• Async Task: Using an async task is the simplest approach, but it is tightly coupled with the activity lifecycle. If the activity dies (or is re-created), any ongoing async task will become wasted cycles or otherwise create unexpected behavior upon returning to the main thread. In addition, it is a terrible idea to drop a response from a network request just because a user rotated his/her phone.
• Loaders: Loaders are a better option, as they recover themselves after a configuration change. On the other hand, they are designed to load data from disk and are not well suited for long-running network requests.
• Service with a Thread Pool: Using a service is a much better solution, as it de-couples business logic from your UI. However, you will need a thread pool (e.g. ThreadPoolExecutor) to process requests in parallel, broadcast events to update the UI, and write additional code to persist queued requests to disk. As your application grows, the number of background operations grows, which force you to consider task prioritization and often-complicated concurrency problems.

Our Solution

Job Queue provides you a nice framework to do all of the above and more. You define your background tasks as [Jobs][11] and enqueue them to your [JobManager][10] instance. Job Manager will take care of prioritization, persistence, load balancing, delaying, network control, grouping etc. It also provides a nice lifecycle for your jobs to provide a better, consistent user experience.

Although not required, it is most useful when used with an event bus. It also supports dependency injection.

• Job Queue was inspired by a [Google I/O 2010 talk on REST client applications][8].

Show me the code

Since a code example is worth thousands of documentation pages, here it is.

File: PostTweetJob.java

…

File: TweetActivity.java

//...
public void onSendClick() {
    final String status = editText.getText().toString();
    if(status.trim().length() > 0) {
      jobManager.addJobInBackground(new PostTweetJob(status));
      editText.setText("");
    }
}
...

That's it. :) Job Manager allows you to enjoy:

• No network calls in activity-bound async tasks
• No serialization mess for important requests
• No "manual" implementation of network connectivity checks or retry logic

Under the hood

• When user clicked the send button, onSendClick() was called, which creates a PostTweetJob and adds it to Job Queue for execution. It runs on a background thread because Job Queue will make a disk access to persist the job.

• Right after PostTweetJob is synchronized to disk, Job Queue calls DependencyInjector (if provided) which will inject fields into our job instance. At PostTweetJob.onAdded() callback, we saved PostTweetJob to disk. Since there has been no network access up to this point, the time between clicking the send button and reaching onAdded() is within fracions of a second. This allows the implementation of onAdded() to inform UI to display the newly sent tweet almost instantly, creating a "fast" user experience. Beware, onAdded() is called on the thread job was added.

• When it's time for PostTweetJob to run, Job Queue will call onRun() (and it will only be called if there is an active network connection, as dictated at the job's constructor). By default, Job Queue uses a simple connection utility that checks ConnectivityManager (ensure you have ACCESS_NETWORK_STATE permission in your manifest). You can provide a custom implementation which can add additional checks (e.g. your server stability). You should also provide a NetworkUtil which can notify Job Queue when network is recovered so that Job Queue will avoid a busy loop and decrease # of consumers(default configuration does it for you).

• Job Queue will keep calling onRun() until it succeeds (or reaches a retry limit). If onRun() throws an exception, Job Queue will call shouldReRunOnThrowable() to allow you to handle the exception and decide whether to retry job execution or abort.

• If all retry attempts fail (or when shouldReRunOnThrowable() returns false), Job Queue will call onCancel() to allow you to clean your database, inform the user, etc.

Advantages

• It is very easy to de-couple application logic from your activites, making your code more robust, easy to refactor, and easy to test.
• You don't have to deal with AsyncTask lifecycles. This is true assuming you use an event bus to update your UI (you should). At Path, we use greenrobot's EventBus; however, you can also go with your favorite. (e.g. [Square's Otto] (https://github.com/square/otto))
• Job Queue takes care of prioritizing jobs, checking network connection, running them in parallel, etc. Job prioritization is especially indispensable when you have a resource-heavy app like ours.
• You can delay jobs. This is helpful in cases like sending a GCM token to your server. It is very common to acquire a GCM token and send it to your server when a user logs in to your app, but you don't want it to interfere with critical network operations (e.g. fetching user-facing content).
• You can group jobs to ensure their serial execution, if necessary. For example, assume you have a messaging client and your user sent a bunch of messages when their phone had no network coverage. When creating these SendMessageToNetwork jobs, you can group them by conversation ID. Through this approach, messages in the same conversation will send in the order they were enqueued, while messages between different conversations are still sent in parallel. This lets you effortlessly maximize network utilization and ensure data integrity.
• By default, Job Queue monitors network connectivity (so you don't need to worry about it). When a device is operating offline, jobs that require the network won't run until connectivity is restored. You can even provide a custom NetworkUtil if you need custom logic (e.g. you can create another instance of Job Queue which runs only if there is a wireless connection).
• It is unit tested and mostly documented. You can check our [code coverage report][3] and [Javadoc][4].

Getting Started

We distribute artifacts through maven central repository.

Gradle: compile 'com.path:android-priority-jobqueue:1.1.2'

Maven:

<dependency>
    <groupId>com.path</groupId>
    <artifactId>android-priority-jobqueue</artifactId>
    <version>1.1.2</version>
</dependency>

You can also [download][5] library jar, sources and javadoc from Maven Central.

We highly recommend checking how you can configure job manager and individual jobs.

• [Configure job manager][10]
• [Configure individual jobs][11]
• [Review sample app][6]
• [Review sample configuration][7]

Version History

• 1.1.2 (Feb 18, 2014)
• Report exceptions to logger if addInBackground fails. (#31)
• 1.1.1 (Feb 8, 2014)
• Fixed an important bug (#35) where jobs in the same group may run in parallel if many of them become available at the same time while multiple consumer threads are waiting for a new job.
• 1.1 (Jan 30, 2014)
• Job Status query API (#18)
• Fixed a stackoverflow bug when network status changes after a long time. (#21)
• 1.0 (Jan 14, 2014):
• Added [parameterized][12] constructor for Job for more readable code.
• Deprecated BaseJob in favor of a more complete Job class.
• 0.9.9 (Dec 16, 2013):
• First public release.

[Wiki][9]

Dependencies

• Job Queue does not depend on any other libraries other than Android SDK.
• For testing, we use:
• • Junit 4 (license)
• • Robolectric (license)
• • Fest Util (license)
• • Hamcrest (license)
• For code coverage report, we use:
• • Cobertura (license)
• Sample Twitter client uses:
• • Twitter4j
• • EventBus
• • Path's fork of greenDAO . (original repo)

Building

We are in the process of moving build system from ant to gradle. Right now, you can build with gradle but if you want to run tests, you'll need ant.

• Clone the repo
• > cd jobqueue
• > ant clean build-jar

This will create a jar file under release folder.

Running Tests

• cd jobqueue

• ant clean test

License

Android Priority Jobqueue is made available under the MIT license:

The MIT License (MIT)

Copyright (c) 2013 Path, Inc.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.