Category Archives: Android Developers Blog

An Open Handset Alliance Project

Alternative input methods for Android TV

Posted by Benjamin Baxter, Developer Advocate and Bacon Connoisseur

Hero image displaying phones and tvs communicating to each other

All TVs have the same problem with keyboard input: It is very cumbersome to hunt and peck for each letter using a D-pad with a remote. And if you make a mistake, trying to correct it compounds the problem.

APIs like Smart Lock and Autofill, can ease user's frustrations, but for certain types of input, like login, you need to collect complex input that is difficult for users using the on-screen keyboard.

With the Nearby Connections API, you can use a second screen to gather input from the user with less friction.

How Nearby Connections works

From the documentation:

"Nearby Connections is an offline peer-to-peer socket model for communication based on advertising and discovering devices in proximity.

Usage of the API falls into two phases: pre-connection, and post-connection.

In the pre-connection phase, Advertisers advertise themselves, while Discoverers discover nearby Advertisers and send connection requests. A connection request from a Discoverer to an Advertiser initiates a symmetric authentication flow that results in both sides independently accepting (or rejecting) the connection request.

After a connection request is accepted by both sides, the connection is established and the devices enter the post-connection phase, during which both sides can exchange data."

In most cases the TV is the advertiser and the phone is the discoverer. In the example below, the assumed second device is a phone. The API and patterns described in this article are not limited to a phone. For example, a tablet could also be the second screen device.

The TV is the Advertiser and the phone is the Discoverer.

Login Example

There are many times when keyboard input is required. Authenticating users and collecting billing information (like zip codes and name on card) are common cases. This example handles a login flow that uses a second screen to see how Nearby Connections can help reduce friction.

1. The user opens your app on her TV and needs to login. You can show a screen of options similar to the setup flow for a new TV.

Android TV setup with prompt to continue on the user's phone.

2. After the user chooses to login with their phone, the TV should start advertising and send the user to the associated login app on their phone, which should start discovering.

There are a variety of solutions to open the app on the phone. As an example, Android TV's setup flow has the user open the corresponding app on their mobile device. Initiating the hand-off is a more a UX concern than a technology concern.

Animation showing setup hand off from TV to phone.

3. The phone app should display the advertising TV and prompt the user to initiate the connection. After the (encrypted -- see Security Considerations below for more on this) connection is established the TV can stop advertising and the phone can stop discovering.

"Advertising/Discovery using Nearby Connections for hours on end can affect a device's battery. While this is not usually an issue for a plugged-in TV, it can be for mobile devices, so be conscious about stopping advertising and discovery once they're no longer needed."

4. Next, the phone can start collecting the user's input. Once the user enters their login information, the phone should send it to the TV in a BYTES payload over the secure connection.

5. When the TV receives the message it should send an ACK (using a BYTES payload) back to the phone to confirm delivery.

6. When the phone receives the ACK, it can safely close the connection.

The following diagram summarizes the sequence of events:

Sequence diagram of order of events to setup a connect and send a message.

UX considerations

Nearby Connections needs location permissions to be able to discover nearby devices. Be transparent with your users. Tell them why they need to grant the location permission on their phone.

Since the TV is advertising, it does not need location permissions.

Start advertising: The TV code

After the user chooses to login on the phone, the TV should start advertising. This is a very simple process with the Nearby API.

override fun onGuidedActionClicked(action: GuidedAction?) {
    super.onGuidedActionClicked(action)
    if( action == loginAction ) {
        // Update the UI so the user knows to check their phone
        navigationFlowCallback.navigateToConnectionDialog()
        doStartAdvertising(requireContext()) { payload ->
            handlePayload(payload)
        }
    }
}

When the user clicks a button, update the UI to tell them to look at their phone to continue. Be sure to offer a way to cancel the remote login and try manually with the cumbersome onscreen keyboard.

This example uses a GuidedStepFragment but the same UX pattern applies to whatever design you choose.

Advertising is straightforward. You need to supply a name, a service id (typically the package name), and a `ConnectionLifeCycleCallback`.

You also need to choose a strategy that both the TV and the phone use. Since it is possible that the users has multiple TVs (living room, bedroom, etc) the best strategy to use is P2P_CLUSTER.

Then start advertising. The onSuccessListener and onFailureListener tell you whether or not the device was able to start advertising, they do not indicate a device has been discovered.

fun doStartAdvertising(context: Context) {
    Nearby.getConnectionsClient(context).startAdvertising(
        context.getString(R.string.tv_name),
        context.packageName,
        connectionLifecycleCallback,
        AdvertisingOptions.Builder().setStrategy(Strategy.P2P_CLUSTER).build()
    )
    .addOnSuccessListener {
        Log.d(LoginStepFragment.TAG, "We are advertising!")
    }
    .addOnFailureListener {
        Log.d(LoginStepFragment.TAG, "We cannot start advertising.")
        Toast.makeText(
            context, "We cannot start advertising.", Toast.LENGTH_LONG)
                .show()
    }
}

The real magic happens in the `connectionLifecycleCallback` that is triggered when devices start to initiate a connection. The TV should accept the handshake from the phone (after performing the necessary authentication -- see Security Considerations below for more) and supply a payload listener.

val connectionLifecycleCallback = object : ConnectionLifecycleCallback() {

    override fun onConnectionInitiated(
            endpointId: String, 
            connectionInfo: ConnectionInfo
    ) {
        Log.d(TAG, "Connection initialized to endpoint: $endpointId")
        // Make sure to authenticate using `connectionInfo.authenticationToken` 
        // before accepting
        Nearby.getConnectionsClient(context)
            .acceptConnection(endpointId, payloadCallback)
    }

    override fun onConnectionResult(
        endpointId: String, 
        connectionResolution: ConnectionResolution
    ) {
        Log.d(TAG, "Received result from connection: ${connectionResolution.status.statusCode}")
        doStopAdvertising()
        when (connectionResolution.status.statusCode) {
            ConnectionsStatusCodes.STATUS_OK -> {
                Log.d(TAG, "Connected to endpoint: $endpointId")
                otherDeviceEndpointId = endpointId
            }
            else -> {
                otherDeviceEndpointId = null
            }
        }
    }

    override fun onDisconnected(endpointId: String) {
        Log.d(TAG, "Disconnected from endpoint: $endpointId")
        otherDeviceEndpointId = null
    }
}

The payloadCallback listens for the phone to send the login information needed. After receiving the login information, the connection is no longer needed. We go into more detail later in the Ending the Conversation section.

Discovering the big screen: The phone code

Nearby Connections does not require the user's consent. However, the location permission must be granted in order for discovery with Nearby Connections to work its magic. (It uses BLE scanning under the covers.)

After opening the app on the phone, start by prompting the user for location permission if not already granted on devices running Marshmallow and higher.

Once the permission is granted, start discovering, confirm the connection, collect the credentials, and send a message to the TV app.

Discovering is as simple as advertising. You need a service id (typically the package name -- this should be the same on the Discoverer and Advertiser for them to see each other), a name, and a `EndpointDiscoveryCallback`. Similar to the TV code, the flow is triggered by callbacks based on the connection status.

Nearby.getConnectionsClient(context).startDiscovery(
        context.packageName,
        mobileEndpointDiscoveryCallback,
        DiscoveryOptions.Builder().setStrategy(Strategy.P2P_CLUSTER).build()
        )
        .addOnSuccessListener {
            // We're discovering!
            Log.d(TAG, "We are discovering!")
        }
         .addOnFailureListener {
            // We were unable to start discovering.
            Log.d(TAG, "We cannot start discovering!")
        }

The Discoverer's listeners are similar to the Advertiser's success and failure listeners; they signal if the request to start discovery was successful or not.

Once you discover an advertiser, the `EndpointDiscoveryCallback` is triggered. You need to keep track of the other endpoint to know who to send the payload, e.g.: the user's credentials, to later.

val mobileEndpointDiscoveryCallback = object : EndpointDiscoveryCallback() {
    override fun onEndpointFound(
        endpointId: String, 
        discoveredEndpointInfo: DiscoveredEndpointInfo
    ) {
        // An endpoint was found!
        Log.d(TAG, "An endpoint was found, ${discoveredEndpointInfo.endpointName}")
        Nearby.getConnectionsClient(context)
            .requestConnection(
                context.getString(R.string.phone_name), 
                endpointId, 
                connectionLifecycleCallback)
    }

    override fun onEndpointLost(endpointId: String) {
        // A previously discovered endpoint has gone away.
        Log.d(TAG, "An endpoint was lost, $endpointId")
    }
}

One of the devices must initiate the connection. Since the Discoverer has a callback for endpoint discovery, it makes sense for the phone to request the connection to the TV.

The phone asks for a connection supplying a `connectionLifecycleCallback` which is symmetric to the callback in the TV code.

val connectionLifecycleCallback = object : ConnectionLifecycleCallback() {
    override fun onConnectionInitiated(
        endpointId: String,
        connectionInfo: ConnectionInfo
    ) {
        Log.d(TAG, "Connection initialized to endpoint: $endpointId")
        // Make sure to authenticate using `connectionInfo.authenticationToken` before accepting
        Nearby.getConnectionsClient(context)
                .acceptConnection(endpointId, payloadCallback)
    }

    override fun onConnectionResult(
        endpointId: String,
        connectionResolution: ConnectionResolution
    ) {
        Log.d(TAG, "Connection result from endpoint: $endpointId")
        when (connectionResolution.status.statusCode) {
            ConnectionsStatusCodes.STATUS_OK -> {
                Log.d(TAG, "Connected to endpoint: $endpointId")
                otherDeviceEndpointId = endpointId
                waitingIndicator.visibility = View.GONE
                emailInput.editText?.isEnabled = true
                passwordInput.editText?.isEnabled = true

                Nearby.getConnectionsClient(this).stopDiscovery()
            }
            else -> {
                otherDeviceEndpointId = null
            }
        }
    }

    override fun onDisconnected(endpointId: String) {
        Log.d(TAG, "Disconnected from endpoint: $endpointId")
        otherDeviceEndpointId = null
    }
}

Once the connection is established, stop discovery to avoid keeping this battery-intensive operation running longer than needed. The example stops discovery after the connection is established, but it is possible for a user to leave the activity before that happens. Be sure to stop the discovery/advertising in onStop() on both the TV and phone.


override fun onStop() {
    super.onStop()
    Nearby.getConnectionsClient(this).stopDiscovery()
}


Just like a TV app, when you accept the connection you supply a payload callback. The callback listens for messages from the TV app such as the ACK described above to clean up the connection.

After the devices are connected, the user can use the keyboard and send their authentication information to the TV by calling `sendPayload()`.

fun sendCreditials() {

    val email = emailInput.editText?.text.toString()
    val password = passwordInput.editText?.text.toString()

    val creds = "$email:$password"
    val payload = Payload.fromBytes(creds.toByteArray())
    Log.d(TAG, "sending payload: $creds")
    if (otherDeviceEndpointId != null) {
        Nearby.getConnectionsClient(this)
                .sendPayload(otherDeviceEndpointId, payload)
    }
}

Ending the conversation

After the phone sends the payload to the TV (and the login is successful), there is no reason for the devices to remain connected. The TV can initiate the disconnection with a simple shutdown protocol.

The TV should send an ACK to the phone after it receives the credential payload.

val payloadCallback = object : PayloadCallback() {
    override fun onPayloadReceived(endpointId: String, payload: Payload) {
        if (payload.type == Payload.Type.BYTES) {
            payload.asBytes()?.let {
                val body = String(it)
                Log.d(TAG, "A payload was received: $body")
                // Validate that this payload contains the login credentials, and process them.

                val ack = Payload.fromBytes(ACK_PAYLOAD.toByteArray())
                Nearby.getConnectionsClient(context).sendPayload(endpointId, ack)
            }
        }
    }

    override fun onPayloadTransferUpdate(
        endpointId: String,
        update: PayloadTransferUpdate
    ) {    }
}

The phone should have a `PayloadCallback` that initiates a disconnection in response to the ACK. This is also a good time to reset the UI to show an authenticated state.

private val payloadCallback = object : PayloadCallback() {
    override fun onPayloadReceived(endpointId: String, payload: Payload) {
        if (payload.type == Payload.Type.BYTES) {
            payload.asBytes()?.let {
                val body = String(it)
                Log.d(TAG, "A payload was received: $body")

                if (body == ACK_PAYLOAD) {
                    waitingIndicator.visibility = View.VISIBLE
                    waitingIndicator.text = getString(R.string.login_successful)
                    emailInput.editText?.isEnabled = false
                    passwordInput.editText?.isEnabled = false
                    loginButton.isEnabled = false

                    Nearby.getConnectionsClient(this@MainActivity)
                        .disconnectFromEndpoint(endpointId)
                }
            }
        }
    }

    override fun onPayloadTransferUpdate(
        endpointId: String,
        update: PayloadTransferUpdate
    ) {    }
}

Security considerations

For security (especially since we're sending over sensitive information like login credentials), it's strongly recommended that you authenticate the connection by showing a code and having the user confirm that the two devices being connected are the intended ones -- without this, the connection established by Nearby Connection is encrypted but not authenticated, and that's susceptible to Man-In-The-Middle attacks. The documentation goes into greater detail on how to authenticate a connection.

Let the user accept the connection by displaying a confirmation code on both devices.

Does your app offer a second screen experience?

There are many times when a user needs to supply input to a TV app. The Nearby API provides a way to offload the hardships of an onscreen-dpad-driven keyboard to an easy and familiar phone keyboard.

What use cases do you have where a second screen would simplify your user's life? Leave a comment or send me (@benjamintravels) or Varun (@varunkapoor, Team Lead for Nearby Connections) a tweet to continue the discussion.

Streaming support spec for hearing aids on Android

Posted by Seang Chau, Vice President, Engineering

According to the World Health Organization1, around 466 million people worldwide have disabling hearing loss. This number is expected to increase to 900 million people by the year 2050. Google is working with GN Hearing to create a new open specification for hearing aid streaming support on future versions of Android. Users with hearing loss will be able to connect, pair, and monitor their hearing aids so they can hear their phones loudly and clearly.

Hearing aid users expect a high quality, low latency experience with minimal impact on phone and hearing aid battery life. We've published a new hearing aid spec for Android smartphones: Audio Streaming for Hearing Aids (ASHA) on Bluetooth Low Energy Connection-Oriented Channels. ASHA is designed to have a minimal impact on battery life with low-latency while maintaining a high quality audio experience for users who rely on hearing aids. We look forward to continually evolving the spec to even better meet the needs of our users.

The spec details the pairing and connectivity, network topology, system architecture, and system requirements for implementing hearing aids using low energy connection-oriented channels. Any hearing aid manufacturer can now build native hearing aid support for Android.

The protocol specification is available here.

Updating Wear OS Google Play Store policy to increase app quality

Posted by Hoi Lam, Lead Developer Advocate, Wear OS by Google

Today we are announcing a new initiative to improve Wear app quality and their presentation in the Google Play Store. The Wear app review process, which has been in place since the launch of Android Wear 2.0, is currently optional. It will become mandatory for apps to be listed on the Wear OS by Google version of the Google Play Store from the following dates:

  • New Wear apps: 1 October 2018
  • Existing Wear apps: 4 March 2019.

The review process for mobile apps remains unchanged, and is independent of the Wear app review. Mobile app updates will not be blocked if they fail the Wear app review.

We hope this lightweight app review process will improve the quality of Wear app experiences across the wide range of devices available to your users. In addition, since screenshots are required for the Wear app review, this will improve the discovery and presentation of your Wear apps in the Google Play Store.

See a comprehensive list of review criteria here. The following are common issues we see during Wear app reviews:

  • Support for different screen types - Wear OS by Google is available in both round and square screens, and some round devices also have a chin. Developers are advised to test on all screen types. If a physical device is unavailable, please use the Wear OS by Google emulator.
  • Wear OS by Google app screenshot - To pass the review, the app needs to have at least one Wear OS app screenshot. To keep pre-release Wear apps private, the Google Play Store will not show the Wear screenshots unless the Wear App is in production or open testing. Currently, the Google Play Store only supports uploading one set of screenshots across all production and test versions. For existing Wear apps, we recommend developers keeping their production Wear app screenshots unchanged when uploading new open test or closed test Wear apps.

Opting out of app review for early prototypes

We understand that some developers need to experiment with their Wear apps in the early stages of app development, and a Wear app review at this stage might not be appropriate. In this case, developers have two options:

Please note that the open test and closed test channels will be subject to Wear app review to help front-load the quality assurance process and to avoid leaving reviews to the last minute.

Thank you for your continuing support of Wear OS by Google.

Google releases source for Google I/O 2018 for Android

Posted by Shailen Tuli, DPE

Today we're releasing the source code for the official Google I/O 2018 for Android app.

The 2018 version constitutes a comprehensive rewrite of the app. For many years, the app has used a ContentProvider + SyncAdapter architecture. This year, we rewrote the app using Architecture Components and brought the code in sync with the Android team's current recommendations for building modern apps.

Architecture

We followed the recommendations laid out in the Guide to App Architecture for writing modular, testable and maintainable code when deciding on the architecture for the app. We kept logic away from Activities and Fragments and moved it to ViewModels. We observed data using LiveData and used the Data Binding Library to bind UI components in layouts to the app's data sources.

The overall architecture of the app can be summarized in this diagram:

We used a Repository layer for handling data operations. IOSched's data comes from a few different sources — user data is stored in Cloud Firestore (either remotely or in a local cache for offline use), user preferences and settings are stored in SharedPreferences, conference data is stored remotely and is fetched and stored in memory for the app to use — and the repository modules are responsible for handling all data operations and abstracting the data sources from the rest of the app. If we ever wanted to swap out the Firestore backend for a different data source in the future, our architecture allows us to do so in a clean way.

We implemented a lightweight domain layer, which sits between the data layer and the presentation layer, and handles discrete pieces of business logic off the UI thread. Examples.

We used Dagger2 for dependency injection and we heavily relied on dagger-android to abstract away boilerplate code.

We used Espresso for basic instrumentation tests and JUnit and Mockito for unit testing.

Firebase

The use of Firebase technologies has grown in the app as the Firebase platform has matured. The 2018 version uses the following Firebase components:

  • Cloud Firestore is our source for all user data (events starred or reserved by a user). Firestore gave us automatic sync and also seamlessly managed offline functionality for us.
  • Firebase Cloud Functions allowed us to run backend code. The reservations feature heavily depended on Functions checking a user's status (only attendees were allowed to make reservations), checking space availability and persisting reservation status in Firestore.
  • Firebase Cloud Messaging let us inform the app about changes to conference data on our server. Conference data is mostly static, but it does change from time to time, especially after the keynote. The app has traditionally used a ping-and-fetch model when working with conference data, and we retained that usage this year.
  • Remote Config helped us manage in-app constants. In previous years, we had found ourselves unable to inform users when data not directly related to the conference schedule — WiFi information, conference shuttle schedule, etc. — changed unexpectedly. Remote Config helped us update such values in a lightweight manner.

Kotlin

We made an early decision to rewrite the app from scratch to bring it in line with modern Android architecture. Using Kotlin for the rewrite was an easy choice: we loved Kotlin's expressive, concise, and powerful syntax; we found that Kotlin's support for safety features including nullability and immutability made our code more resilient; and we leveraged the enhanced functionality provided by Android Ktx extensions.

Material Design

At I/O 2018, the Material Design team announced Material Theming, giving apps much greater ability to customize Material Design to bring more of their product's brand. As we launched the app before Material Theming, we couldn't use all of the new components but we managed to sneak a couple in like the new Bottom App Bar with inset Floating Action Button and we were able to incorporate a lot of the conference's branding elements.

Future plans

The rewrite of the app brings the code in sync with Android's opinionated recommendations about building apps, and it resulted in a cleaner, more maintainable codebase. We'll continue working on the app, incorporating JetPack components as they become available and finding opportunities to showcase platform features that are good fits for the app. Developers can follow changes to the code on GitHub.

Looking forward with Google Play

Posted by Purnima Kochikar, Director, Google Play, Apps & Games

On Monday we released Android 9 Pie. As we continue to push the Android platform forward, we're always looking to provide new ways to distribute your apps efficiently, help people discover and engage with your work, and improve the overall security of our ecosystem. Google Play has had a busy year so far with some big milestones around helping you reach more users, including:

  • Shrinking download size: Android App Bundle & Dynamic Delivery has helped reduce app sizes by up to 65%, leading to increased downloads and fewer uninstalls.
  • Helpling improve quality: New tools in the Play Console have helped you reduce crash rates by up to 70%.
  • Improving discovery: Improvements to the discovery experience has increased Google Play Store visits by 30% over the last 12 months.
  • Keeping users safe: Google Play Protect scans more than 50 billion apps a day and Android API level 26 adoption requirements improve app security and performance.

Google Play is dedicated to helping you build and grow quality app businesses, reach the more than 2 billion Android devices globally and provide your users with better experiences. Here are some of the important areas we're prioritizing this year:

Innovative Distribution

We've added more testing tools to the popular Play Console to help developers de-risk app launches with internal and external test tracks and staged rollouts to get valuable early feedback. This year we've expanded the Start on Android program globally that provides developers new to Android additional guidance to optimize their apps before launch. Google Play Instant remains a huge bet to transform app discovery and improve conversions by letting users engage without the friction of installing. We're seeing great results from early adopters and are working on new places to surface instant experience, including ads, and making them easier to build throughout the year.

Improving App Quality

Google Play plays an important role helping developers understand and fix quality and performance issues. At I/O, we showcased how we expanded the battery, stability and rendering of Android vitals reporting to include app start time & permission denials, enabling developers to cut application not responding errors by up to 95%. We also expanded the functionality of automated device testing with the pre-launch report to enable games testing. Recently, we increased the importance of app quality in our search and discovery recommendations that has resulted in higher engagement and satisfaction with downloaded games.

Richer Discovery

Over the last year we've rolled out more editorial content and improved our machine learning to deliver personalized recommendations for apps and games that engage users. Since most game downloads come from browsing (as opposed to searching or deep linking into) the store, we've put particular focus on games discovery, with a new games home page, special sections for premium and new games, immersive video trailers and screenshots, and the ability to try games instantly. We've also introduced new programs to help drive app downloads through richer discovery. For example, since launching our app pre-registration program in 2016, we've seen nearly 250 million app pre-registrations. Going forward, we'll be expanding on these programs and others like LiveOps cards to help developers engage more deeply with their audience.

Expanding Commerce Platform

Google Play now collects payments in 150 markets via credit card, direct carrier billing (DCB), Paypal, and gift cards. Direct carrier billing is now enabled across 167 carriers in 64 markets. In 2018, we have focused on expanding our footprint in Africa and Latam with launches in Ghana, Kenya, Tanzania, Nigeria, Peru & Colombia. And users can now buy Google Play credit via gift cards or other means in more 800,000 retail locations around the world. This year, we also launched seller support in 18 new markets bringing the total markets with seller support to 98. Our subscription offering continues to improve with ML-powered fraud detection and even more control for subscribers and developers. Google Play's risk modeling automatically helps detect fraudulent transactions and purchase APIs help you better analyze your refund data to identify suspicious activity.

Maintaining a Safe & Secure Ecosystem

Google Play Protect and our other systems scan and analyze more than 50 billion apps a day to keep our ecosystem safe for users and developers. In fact, people who only download apps from Google Play are nine times less likely to download a potentially harmful app than those who download from other sources. We've made significant improvements in our ability to detect abuse—such as impersonation, inappropriate content, fraud, or malware—through new machine learning models and techniques. The result is that 99% of apps with abusive content are identified and rejected before anyone can install them. We're also continuing to run the Google Play Security Rewards Program through a collaboration with Hacker One to discover other vulnerabilities.

We are continually inspired by what developers build—check out #IMakeApps for incredible examples—and want every developer to have the tools needed to succeed. We can't wait to see what you do next!

Meet the first Indie Games Accelerator class

Posted by Vineet Tanwar, Business Development Manager, Google Play

In June, we announced the Indie Games Accelerator, a new four month program to help indie game startups from India, Pakistan and Southeast Asia supercharge their growth on Android. We have been truly impressed by the overwhelming responses we have received, and the creativity that indie game developers from these regions have to offer.

We had a great time going through the applications and playing the games which were submitted for review. Now, it's finally time to announce the inaugural class of startups selected for the program who we will mentor and coach over the next few months. Here they are:

Congratulations to the selected participants and a huge thanks to everyone that applied! Find out more about the program or express your interest in joining next class of Indie Games Accelerator.

How useful did you find this blogpost?

Android Pie SDK is now more Kotlin-friendly

Posted by James Lau, Product Manager (@jmslau)

When using the Java programming language, one of the most common pitfalls is trying to access a member of a null reference, causing a NullPointerException to be thrown. Kotlin offers protection against this by baking nullable and non-nullable types into the type system. This helps eliminate NullPointerExceptions from your code and improve your app's overall quality. When Kotlin code is calling into APIs written in the Java programming language, it relies on nullability annotations in those APIs to determine the nullability of each parameter and the return type. Unannotated parameters and return types are treated as platform types, which weakens the null-safety guarantee of Kotlin.

As part of yesterday's Android 9 announcement, we have also released a new Android SDK that contains nullability annotations for some of the most frequently used APIs. This will preserve the null-safety guarantee when your Kotlin code is calling into any annotated APIs in the SDK. Even if you are using the Java programming language, you can still benefit from these annotations by using Android Studio to catch nullability contract violations.

Not a breaking change

Normally, nullability contract violations in Kotlin result in compilation errors. But to ensure the newly annotated APIs are compatible with your existing code, we are using an internal mechanism provided by the Kotlin compiler team to mark the APIs as recently annotated. Recently annotated APIs will result only in warnings instead of errors from the Kotlin compiler. You will need to use Kotlin 1.2.60 or later.

Our plan is to have newly added nullability annotations produce warnings only, and increase the severity level to errors starting in the following year's Android SDK. The goal is to provide you with sufficient time to update your code.

How to use the "Kotlin-friendly" SDK

To get started, go to Tools > SDK Manager in Android Studio. Select Android SDK on the left menu, and make sure the SDK Platforms tab is open.

Use SDK Manager in Android Studio to install SDK for API Level 28 Revision 6

Check Android 8.+ (P) and click OK. This will install the Android SDK Platform 28 revision 6 if it is not already installed. After that, set your project's compile SDK version to API 28 to start using the new Android Pie SDK with nullability annotations.

Use the Project Structure Dialog to change your project's Compile Sdk Version to API 28

You may also need to update your Kotlin plugin in Android Studio if it's not already up-to-date. Make sure your Kotlin plugin version is 1.2.60 or later by going to Tools > Kotlin > Configure Kotlin Plugin Updates.

Once it's set up, your builds will start showing warnings if you have any code that violates nullability contracts in the Android SDK. An example of such a warning is shown below.

Sample warning from the Kotlin compiler when code violates a recently added nullability contract in the Android SDK.

You will also start seeing warnings in Android Studio's code editor if you call an Android API with the incorrect nullability. An example is shown below.

Android Studio warning about passing a null reference to a parameter annotated as a recently non-null type in the android.graphics.Path API.

Leveraging nullability annotations from the Java programming language

You can benefit from the new nullability annotations even if your code is in the Java programming language. By default, Android Studio will highlight any nullability contract violations with a warning, like the one below:

Android Studio showing a warning about nullability contract violation in code written in the Java programming language

To ensure that you have this inspection enabled, you can go to the IDE's settings page and search for "Constant conditions & exceptions" inspection and make sure that item is checked.

Use the Inspections page under Settings to ensure the Constant conditions & exceptions code inspection is enabled.

If you are using the Java programming language, nullability contract violations will not produce any compiler warning or error. Only the in-IDE code inspections are available to flag these issues.

You can also run code inspections across your entire project and see the aggregated results. Click on Analyze > Inspect Code… to start.

What's Next

The Android SDK API surface is very large, and we have only annotated a small percentage of the APIs so far - there is still lots of work remaining. Over the next several Android SDK releases, we will continue to add nullability annotations to the existing Android APIs, as well as making sure new APIs are annotated.

With the "Kotlin-friendly" Android SDK, the nullability annotations in AndroidX (part of the Jetpack family), and Android KTX, we are continuing to improve the Android APIs for developers using Kotlin. If you have not yet tried Kotlin, we encourage you to try it. Not only can Kotlin make your code more concise, it can also improve the stability of your apps.

Happy Kotlin-ing!

Introducing Android 9 Pie

Posted by Dave Burke, VP of Engineering

After more than a year of development and months of testing by early adopters, we're ready to launch Android 9 Pie, the latest release of Android, to the world.

Android 9 harnesses the power of machine learning to make your phone smarter, simpler, and tailored to you. Read all about the new consumer features here. For developers, Android 9 includes many new ways to enhance your apps and build new experiences to drive engagement.

You've given us tons of feedback along the way--over a thousand bugs and feature requests--thank you! More than 140,000 of you tried our preview builds through the Android Beta program, and seven of our device maker partners also brought our Beta to their flagship devices, enabling users around the world to give their feedback too.

Today we're pushing the source code to Android Open Source Project (AOSP), and starting the Android 9 rollout to all Pixel users worldwide, with Android 9 coming to many more devices in the coming months.

We continue to move Android forward as the premier open platform for developers worldwide to build their businesses. With Android 9 -- together with the powerful new capabilities in Google Play for apps and games -- we're committed to helping you build great experiences, as well as reach and engage the right users safely and cost-effectively around the world.

What's in Android 9?

A smarter smartphone, with machine learning at the core

Android 9 helps your phone learn as you use it, by picking up on your preferences and adjusting automatically. Everything from helping users get the most out of their battery life to surfacing the best parts of the apps they use all the time, right when they need it most, Android 9 keeps things running smoother, longer.

Adaptive Battery

We partnered with DeepMind on a feature called Adaptive Battery that uses machine learning to prioritize system resources for the apps the user cares about most. If your app is optimized for Doze, App Standby, and Background Limits, Adaptive Battery should work well for you right out of the box. If you haven't yet taken optimized your app, make sure to check out the details in the power documentation to see how it works.

Slices

Slices can help users perform tasks faster by enabling engagement outside of the fullscreen app experience. It does this by using UI templates that can display rich, dynamic, and interactive content from your app from within the Google Search app and later in other places like the Google Assistant. You can learn more about building Slices to enhance your app here.

App Actions

App Actions is a new way to raise the visibility of your app and drive engagement. Actions take advantage of machine learning to surface your app to the user at just the right time, based on your app's semantic intents and the user's context.

We'll be sharing more details in the coming weeks on registering your app to handle one or more user intents, so your apps can be enabled for App Actions and surfaced across multiple Google and Android surfaces in response to user queries.

Text Classifier and Smart Linkify

We've extended the ML models that identify entities in content or text input to support more types like Dates and Flight Numbers through the TextClassifier API. Smart Linkify lets you take advantage of the TextClassifier models through the Linkify API, including enriched options for quick follow-on user actions. Smart Linkify also delivers significant improvements in accuracy of detection as well as performance.

Neural Networks API 1.1

Android 9 adds an updated version of the Neural networks API, to extend Android's support for accelerated on-device machine learning. Neural Networks 1.1 adds support for nine new ops -- Pad, BatchToSpaceND, SpaceToBatchND, Transpose, Strided Slice, Mean, Div, Sub, and Squeeze. A typical way to take advantage of the APIs is through TensorFlow Lite.

Getting the most from your phone -- more easily

We're excited about making your smartphone more intelligent. But it's also important that the technology fades to the back for users. In Android 9, we've evolved Android's UI to be simpler and more approachable -- for developers, these changes help improve the way users find, use, and manage your apps.

New system navigation

Android 9 introduces a new system navigation that we've been working on for more than a year. The new design helps make Android's multitasking more approachable and makes discovering apps much easier. You can swipe up from anywhere to see full-screen previews of recently used apps and simply tap to jump back into one of them.

Display cutout

Now your app can take full advantage of the latest edge-to-edge screens through display cutout support in Android 9. For most apps, supporting display cutout is seamless, with the system managing status bar height to separate your content from the cutout. If you have immersive content, you can use the display cutout APIs to check the position and shape of the cutout and request full-screen layout around it. To help with development and testing, we've added a Developer Option that simulates several cutout shapes on any device.

Apps with immersive content can display content fullscreen on devices with a display cutout.

Notifications and smart reply

Android 9 makes notifications even more useful and more actionable. Messaging apps can take advantage of the new MessagingStyle APIs to show conversations, attach photos and stickers, and even suggest smart replies. You'll soon be able to use ML Kit to generate smart reply suggestions for your app.

MessagingStyle notifications with conversations and smart replies [left], images and stickers [right].

Text Magnifier

In Android 9 we've added a Magnifier widget to improve the user experience of selecting text. The Magnifier widget lets users precisely position the cursor or the text selection handles by viewing zoomed text through a draggable pane. You can attach it to any view that is attached to a window, so you can use it in custom widgets or during custom text-rendering. The Magnifier widget can also provide a zoomed-in version of any view or surface, not just text.

Check out our recent blog post for more about this and other Text features, such as PrecomputedText and line height and baseline text alignment.

Security and privacy for users

Biometric prompt

With a range of biometric sensors in use for authentication, we've made the experience more consistent across sensor types and apps. Android 9 introduces a system-managed dialog to prompt the user for any supported type of biometric authentication. Apps no longer need to build their own dialog--instead they use the BiometricPrompt API to show the standard system dialog. In addition to Fingerprint (including in-display sensors), the API supports Face and Iris authentication.

If your app is drawing its own fingerprint auth dialogs, you should switch to using the BiometricPrompt API as soon as possible. See this post for more information.

Protected Confirmation

Android 9 introduces Android Protected Confirmation, which uses the Trusted Execution Environment (TEE) to guarantee that a given prompt string is shown and confirmed by the user. Only after successful user confirmation will the TEE then sign the prompt string, which the app can verify.

Stronger protection for private keys

We've added StrongBox as a new KeyStore type, providing API support for devices that provide key storage in tamper-resistant hardware with isolated CPU, RAM, and secure flash. You can set whether your keys should be protected by a StrongBox security chip in your KeyGenParameterSpec.

DNS over TLS

Android 9 adds built-in support for DNS over TLS, automatically upgrading DNS queries to TLS if a network's DNS server supports it. Users can manage DNS over TLS behavior in a new Private DNS Mode in Network & internet settings. Apps that perform their own DNS queries can use a new API, LinkProperties.isPrivateDnsActive(), to check the DNS mode. More in this post.

HTTPS by default

As part of a larger effort to move all network traffic away from cleartext (unencrypted HTTP) to websites secured with TLS (HTTPS), we're changing the defaults for Network Security Configuration to block all cleartext traffic. You'll now need to make connections over TLS, unless you explicitly opt-in to cleartext for specific domains. See the details here.

Compiler-based security mitigations

In Android 9 we've expanded our use of compiler-level mitigations to harden the platform through run-time detection of dangerous behavior. Control Flow Integrity (CFI) techniques help to prevent code-reuse attacks and arbitrary code execution. In Android 9 we've greatly expanded CFI usage within the media framework and other security-critical components, such as NFC and Bluetooth. We've also introduced CFI kernel support into the Android common kernel when building with LLVM.

We've also expanded our use of Integer overflow sanitizers to mitigate memory-corruption and information-disclosure vulnerabilities. We've prioritized sanitizers in libraries with past vulnerabilities or where complex untrusted input is processed, such as libui, libnl, libmediaplayerservice and others. See this post for details.

Privacy for users

Android 9 safeguards privacy in a number of new ways. The system now restricts access to mic, camera, and all SensorManager sensors from apps that are idle. While your app's UID is idle, the mic reports empty audio and sensors stop reporting events. Cameras used by your app are disconnected and will generate an error if the app tries to use them. In most cases, these restrictions should not introduce new issues for existing apps, but we recommend removing these requests from your apps.

Android 9 also gives the user control over access to the platform's build.serial identifier by putting it behind the READ_PHONE_STATE permission. To access the build.serial identifier, you should use the Build.getSerial() method.

Read more about all of the privacy changes here.

New experiences in camera, audio, and graphics

Multi-camera API and other camera updates

With Android 9 you can now open streams from two or more physical cameras simultaneously on devices that support the multi-camera API. On devices with either dual-front or dual-back cameras, you can create innovative features not possible with just a single camera, such as seamless zoom, bokeh, and stereo vision. The API also lets you call a logical or fused camera stream that automatically switches between two or more cameras.

Other improvements in camera include new Session parameters that help to reduce delays during initial capture, and Surface sharing that lets camera clients handle various use-cases without the need to stop and start camera streaming. We've also added APIs for display-based flash support and access to OIS timestamps for app-level image stabilization and special effects.

HDR VP9 Video and HEIF image compression

Android 9 adds built-in support for HDR VP9 Profile 2, so you can now deliver HDR-enabled movies to your users on HDR-capable devices.

We're excited to add HEIF (heic) image encoding to the platform. HEIF is a popular format for photos that improves compression to save on storage and network data. With platform support on Android 9 devices, it's easy to send and utilize HEIF images from your backend server. Once you've made sure that your app is compatible with this data format for sharing and display, give HEIF a try as an image storage format in your app. You can do a jpeg-to-heic conversion using ImageDecoder or BitmapFactory to obtain a bitmap from jpeg, and you can use HeifWriter in the AndroidX library to write HEIF still images from YUV byte buffer, Surface, or Bitmap.

Enhanced audio with Dynamics Processing

The Dynamics Processing API lets you use a new audio effect to isolate specific frequencies and lower loud or increase soft sounds to enhance the acoustic quality of your app. For example, you can improve the sound of someone who speaks quietly in a loud, distant or otherwise acoustically challenging environment. The API gives you access to a multi-stage, multi-band dynamics processing effect that includes a pre-equalizer, a multi-band compressor, a post-equalizer and a linked limiter.

ImageDecoder for bitmaps and drawables

An ImageDecoder API gives you an easier way to decode images to bitmaps or drawables. You can create a bitmap or drawable from a byte buffer, file, or URI. The API offers several advantages over BitmapFactory, including support for exact scaling, single-step decoding to hardware memory, support for post-processing in decode, and decoding of animated images. You can read more here.

Connectivity and location

Wi-Fi RTT for indoor positioning

Android 9 lets you build indoor positioning features into your apps through platform support for the IEEE 802.11mc Wi-Fi protocol -- also known as Wi-Fi Round-Trip-Time (RTT). On Android 9 devices with hardware support, location permission, and location enabled, your apps can use RTT APIs to measure the distance to nearby Wi-Fi Access Points (APs). The device doesn't need to connect to the APs to use RTT, and to maintain privacy, only the phone is able to determine the distance, not the APs.

Knowing the distance to 3 or more APs, you can calculate the device position with an accuracy of 1 to 2 meters. With this accuracy you can support use-cases like in-building navigation; fine-grained location-based services such as disambiguated voice control (e.g. 'Turn on this light'); and location-based information (e.g. 'Are there special offers for this product?').

Data cost sensitivity in JobScheduler

JobScheduler is Android's central service to help you manage scheduled tasks or work across Doze, App Standby, and Background Limits. In Android 9, JobScheduler handles network-related jobs better for the user, coordinating with network status signals provided separately by carriers. Jobs can now declare their estimated data size, signal prefetching, and specify detailed network requirements—carriers can report networks as being congested or unmetered. JobScheduler then manages work according to the network status. For example, when a network is congested, JobScheduler might defer large network requests. When unmetered, it can run prefetch jobs to improve the user experience, such as prefetching headlines.

Open Mobile API for NFC payments and secure transactions

Android 9 adds an implementation of the GlobalPlatform Open Mobile API to Android. On supported devices, apps can use the OMAPI API to access secure elements (SE) to enable smart-card payments and other secure services. A hardware abstraction layer (HAL) provides the underlying API for enumerating the variety of Secure Elements (eSE, UICC, and others) available.

Performance for apps

ART performance

Android 9 brings performance and efficiency improvements to all apps through the ART runtime. We've expanded ART's use of execution profiles to optimize apps and reduce the in-memory footprint of compiled app code. ART now uses profile information for on-device rewriting of DEX files, with reductions up to 11% across a range of popular apps. We expect these to correlate closely with reductions in system DEX memory usage and faster startup times for your apps.

Optimized for Kotlin

Kotlin is a first-class language on Android, and if you haven't tried it yet, you should! We've made an enduring commitment to Kotlin in Android and continue to expand support including optimizing the performance of Kotlin code. In Android 9, you'll see the first results of this work--we've improved several compiler optimizations, especially those that target loops, to extract better performance. We're also continuing to work in partnership with JetBrains to optimize Kotlin's generated code. You can get all of the latest Kotlin performance improvements just by keeping Android Studio's Kotlin plugin up-to-date.

Today, we are also releasing an update to the Android 9 - API 28 SDK (rev. 6), which contains nullability annotations in some of the most frequently used APIs. We'll provide more details about this in an upcoming post.

Modern Android

As part of Android 9 we are modernizing the foundations of Android and the apps that run on it, as part of our deep, sustained investments in security, performance, and stability.

As we announced last year, Google Play will require all app updates to target Android Oreo (targetSdkVersion 26 or higher) by November 2018. In line with that, if your app targets a platform earlier than Android 4.2 (API level 17), users installing it will see a warning dialog after that day. Here's a checklist of resources for help and support as you migrate -- we're looking forward to seeing your apps getting the most from modern Android.

Get your apps ready for Android 9!

With Android 9 coming to Pixel users starting today, and to other devices in the months ahead, it's important to test your app for compatibility as soon as possible. Just install your current app from Google Play on a device or or emulator running Android 9. As you work through the flows, make sure your app runs and looks great, and that it handles the Android 9 behavior changes properly.

Also watch for uses of non-SDK interfaces in your app. Android 9 restricts access to selected non-SDK interfaces, so you should reduce your reliance on them. See our recent post for details.

After you've made any necessary updates, we recommend publishing to Google Play right away. without changing the app's platform targeting. This lets you ensure a great experience for Android 9 users while you work on enhancing your app with Android 9 APIs and targeting.

Enhance your app with Android 9 features and APIs

When you're ready, dive into Android 9 and build with the new features and APIs in Android 9.

To get started, just download the official API 28 SDK and the latest tools and emulator images into Android Studio 3.1, or use the latest version of Android Studio 3.2. Then update your project's compileSdkVersion and targetSdkVersion to API 28. When you change your targeting, make sure your app supports all of the applicable behavior changes.

As soon as you're ready, publish your APK updates to Google Play. A common strategy is to use Google Play's beta testing feature to get early feedback from a small group of users and then do a staged rollout to production.

Visit the Android 9 site for details and developer documentation. Also check out this video and the Google I/O Android Playlist for more on what's new in Android 9 for developers.

Coming to a device near you

Starting today, an over-the-air update to Android 9 will begin rolling out to Pixel phones. And devices that participated in the Beta program from Sony Mobile, Xiaomi, HMD Global, Oppo, Vivo, OnePlus, and Essential, as well as all qualifying Android One devices, will receive this update by the end of this fall! We are also working with a number of other partners to launch or upgrade devices to Android 9 this year.

As always, the system images for Pixel devices are available here for manual flash and download. If you're looking for the Android 9 source, you'll find it here in the Android Open Source Project repository under the Android 9 branches.

What's next?

Now that we've reached the official release, we're bringing the Developer Preview to a close. We'll soon be closing the Developer Preview issue tracker to new issues, so if you have feedback, feel free to file a new issue against Android 9 in the AOSP issue tracker.

Thanks again to the many developers and early adopters who participated in the Android 9 Developer Preview and public beta. Your contributions have been critical to making the Android 9 platform a great one for developers and consumers.

Supporting display cutouts on edge-to-edge screens

Posted By Megan Potoski, Product Manager, Android System UI

Smartphones are quickly moving towards smaller bezels and larger aspect ratios. On these devices, display cutouts are a popular way to achieve an edge-to-edge experience while providing space for important sensors on the front of the device. There are currently 16 cutout devices from 11 OEMs already released, including several Android P beta devices, with more on the way.

These striking displays present a great opportunity for you to showcase your app. They also mean it's more important than ever to make sure your app provides a consistently great experience across devices with one or two display cutouts, as well as devices with 18:9 and larger aspect ratios.

Examples of cutout devices: Essential PH-1 (left) and Huawei P20 (right).

Make your app compatible with display cutouts

With many popular and upcoming devices featuring display cutouts, what can you do to make sure your app is cutout-ready?

The good news is, for the most part your app should work as intended even on a cutout device. By default, in portrait mode with no special flags set, the status bar will be resized to be at least as tall as the cutout and your content will display in the window below. In landscape or fullscreen mode, your app window will be letterboxed so that none of your content is displayed in the cutout area.

However, there are a few areas where your app could have issues on cutout devices.

  • Watch out for any sort of hard-coding of status bar height -- this will likely cause problems. If possible, use WindowInsetsCompat to get status bar height.
  • In fullscreen, be careful to consider when to use window vs. screen coordinates, as your app will not take up the whole screen when letterboxed. For example, if you use MotionEvent.getRawX/Y() to get screen coordinates for touch events, make sure to transform them to the view's coordinates using getLocationOnScreen().
  • Pay special attention to transitions in and out of fullscreen mode.

Here are a few guidelines describing what issues to look out for and how to fix them.

Take advantage of the cutout area

Rendering your app content in the cutout area can be a great way to provide a more immersive, edge-to-edge experience for users, especially for content like videos, photos, maps, and games.

An example of an app that has requested layout in the display cutout.

In Android P we added APIs to let you manage how your app uses the display cutout area, as well as to check for the presence of cutouts and get their positions.

You can use layoutInDisplayCutoutMode, a new window layout mode, to control how your content is displayed relative to the cutout. By default, the app's window is allowed to extend into the cutout area if the cutout is fully contained within a system bar. Otherwise, the window is laid out such that it does not overlap with the cutout. You can also set layoutInDisplayCutoutMode to always or never render into the cutout. Using SHORT_EDGES mode to always render into the cutout is a great option if you want to take advantage of the full display and don't mind if a bit of content gets obscured by the cutout.

If you are rendering into the cutout, you can use getDisplayCutout() to retrieve a DisplayCutout that has the cutout's safe insets and bounding box(es). These let you check whether your content overlaps the cutout and reposition things if needed.

<style name="ActivityTheme">
  <item name="android:windowLayoutInDisplayCutoutMode">
    default/shortEdges/never
  </item>
</style>

Attribute for setting layoutInDisplayCutoutMode from the Activity's theme.

For devices running Android 8.1 (API 27), we've also back-ported the layoutInDisplayCutoutMode activity theme attribute so you can control the display of your content in the cutout area. Note that support on devices running Android 8.1 or lower is up to the device manufacturer, however.

To make it easier to manage your cutout implementation across API levels, we've also added DisplayCutoutCompat in the AndroidX library, which is now available through the SDK manager.

For more about the display cutout APIs, take a look at the documentation.

Test your app with cutout

We strongly recommend testing all screens and experiences of your app to make sure that they work well on cutout devices. We recommend using one of the Android P Beta Devices that features a cutout, such as the Essential PH-1.

If you don't have a device, you can also test using a simulated cutout on any device running Android P or in the Android Emulator. This should help you uncover any issues that your app may run into on devices with cutouts, whether they are running Android 8.1 or Android P.

What to expect on devices with display cutouts

Android P introduces official platform support for display cutouts, with APIs that you can use to show your content inside or outside of the cutout. To ensure consistency and app compatibility, we're working with our device manufacturer partners to mandate a few requirements.

First, devices must ensure that their cutouts do not negatively affect apps. There are two key requirements:

  • In portrait orientation, with no special flags set, the status bar must extend to at least the height of the cutout.
  • In fullscreen or landscape orientation, the entire cutout area must be letterboxed.

Second, devices may only have up to one cutout on each short edge of the device. This means that:

  • You won't see multiple cutouts on a single edge, or more than two cutouts on a device.
  • You won't see a cutout on the left or right long edge of the device.

Within these constraints, devices can place cutouts wherever they want.

Special mode

Some devices running Android 8.1 (API level 27) or earlier may optionally support a "special mode" that lets users extend a letterboxed fullscreen or landscape app into the cutout area. Devices would typically offer this mode through a toggle in the navigation bar, which would then bring up a confirmation dialog before extending the screen.

Devices that offer "special mode" allow users to optionally extend apps into the cutout area if supported by the app.

If your app's targetSdkVersion is 27 or higher, you can set the layoutInDisplayCutoutMode activity theme attribute to opt-out of special mode if needed.

Don't forget: larger aspect ratios too!

While you are working on cutout support, it's also a great time to make sure your app works well on devices with 18:9 or larger aspect ratios, especially since these devices are becoming increasingly common and can feature display cutouts.

We highly encourage you to support flexible aspect ratios so that your app can leverage the full display area, no matter what device it's on. You should test your app on different display ratios to make sure it functions properly and looks good.

Here are some guidelines on screens support to keep in mind as you are developing, also refer to our earlier post on larger aspect ratios for tips on optimizing. If your app can't adapt to the aspect ratios on long screens, you can choose to declare a max aspect ratio to request letterboxing on those screens.

Thanks for reading, and we hope this helps you deliver a delightful experience to all your users, whatever display they may have!

AndroidX Development is Now Even Better

Posted by Aurimas Liutikas, software engineer on AndroidX team

AndroidX (previously known as Android Support Library) started out as a small set of libraries intended to provide backwards compatibility for new Android platform APIs and, as such, its development was strictly tied to the platform. As a result, all work was done in internal Google branches and then pushed to the public Android Open Source Project (AOSP) together with the platform push. With this flow, external contributions were limited to a narrow window of time where the internal and AOSP branches were close in content. On top of that, it was difficult to contribute -- in order to do a full AndroidX build and testing, external developers had to check out >40GB of the full Android platform code.

Today, the scope of AndroidX has expanded dramatically and includes libraries such as AppCompat for easier UI development, Room for database management, and WorkManager for background work. Many of these libraries implement higher-level abstractions and are less tied to new revisions of the Android platform, and all libraries are designed with backwards compatibility in mind from the start. Several libraries, such as RecyclerView and Fragment, are purely AndroidX-side implementations with few ties to the platform.

Starting a little over two years ago, we began a process of unbundling -- moving AndroidX out of Android platform builds into its own separate build. We had to do a great deal of work, including migrating our builds from make to Gradle as well as migrating all of our API tracking tools and documentation generation out of the platform build. With that process completed, we reached a point where a developer can now check out a minimal AndroidX project, open it in Android Studio, and build using the public SDK and public Android Gradle Plugin.

The Android developer community has long expressed a desire to contribute more easily to AndroidX; however, this was always a challenge due to the reasons described above. This changes today: AndroidX development is moving to public AOSP. That means that our primary feature development (except for top-secret integrations with the platform 😀) and bug fixes will be done in the open using the r.android.com Gerrit review tool and changes will be visible in the aosp/androidx-master-dev branch.

We are making this change to give better transparency to developers; it gives developers a chance to see features and bug fixes implemented in real-time. We are also excited about receiving bug fix contributions from the community. We have written up a short guide on how to go about contributing a patch.

In addition to regular development, AOSP will be a place for experimentation and prototyping. You will see new libraries show up in this repository; some of them may be removed before they ship, change dramatically during pre-alpha development, or merge into existing libraries. The general rule is that only the libraries on maven.google.com are officially ready for external developer usage.

Finally, we are just getting started. We apologize for any rough edges that you might have when contributing to AndroidX, and we request your feedback via the public AndroidX tracker if you hit any issues.