Wednesday, 12 June 2019

File naming

Class files

Class names are written in UpperCamelCase.
For classes that extend an Android component, the name of the class should end with the name of the component; for example: SignInActivity, SignInFragment, ImageUploaderService, ChangePasswordDialog.

1.2.2 Resources files

Resources file names are written in lowercase_underscore.

1.2.2.1 Drawable files

Naming conventions for drawables:
Asset Type Prefix Example
Action bar ab_ ab_stacked.9.png
Button btn_ btn_send_pressed.9.png
Dialog dialog_ dialog_top.9.png
Divider divider_ divider_horizontal.9.png
Icon ic_ ic_star.png
Menu menu_ menu_submenu_bg.9.png
Notification notification_ notification_bg.9.png
Tabs tab_ tab_pressed.9.png
Naming conventions for icons (taken from Android iconography guidelines):
Asset Type Prefix Example
Icons ic_ ic_star.png
Launcher icons ic_launcher ic_launcher_calendar.png
Menu icons and Action Bar icons ic_menu ic_menu_archive.png
Status bar icons ic_stat_notify ic_stat_notify_msg.png
Tab icons ic_tab ic_tab_recent.png
Dialog icons ic_dialog ic_dialog_info.png
Naming conventions for selector states:
State Suffix Example
Normal _normal btn_order_normal.9.png
Pressed _pressed btn_order_pressed.9.png
Focused _focused btn_order_focused.9.png
Disabled _disabled btn_order_disabled.9.png
Selected _selected btn_order_selected.9.png

1.2.2.2 Layout files

Layout files should match the name of the Android components that they are intended for but moving the top level component name to the beginning. For example, if we are creating a layout for the SignInActivity, the name of the layout file should be activity_sign_in.xml.
Component Class Name Layout Name
Activity UserProfileActivity activity_user_profile.xml
Fragment SignUpFragment fragment_sign_up.xml
Dialog ChangePasswordDialog dialog_change_password.xml
AdapterView item --- item_person.xml
Partial layout --- partial_stats_bar.xml
A slightly different case is when we are creating a layout that is going to be inflated by an Adapter, e.g to populate a ListView. In this case, the name of the layout should start with item_.
Note that there are cases where these rules will not be possible to apply. For example, when creating layout files that are intended to be part of other layouts. In this case you should use the prefix partial_.

1.2.2.3 Menu files

Similar to layout files, menu files should match the name of the component. For example, if we are defining a menu file that is going to be used in the UserActivity, then the name of the file should be activity_user.xml
A good practice is to not include the word menu as part of the name because these files are already located in the menu directory.

1.2.2.4 Values files

Resource files in the values folder should be plural, e.g. strings.xml, styles.xml, colors.xml, dimens.xml, attrs.xml

2 Code guidelines

2.1 Java language rules

2.1.1 Don't ignore exceptions

You must never do the following:
void setServerPort(String value) {
    try {
        serverPort = Integer.parseInt(value);
    } catch (NumberFormatException e) { }
}
While you may think that your code will never encounter this error condition or that it is not important to handle it, ignoring exceptions like above creates mines in your code for someone else to trip over some day. You must handle every Exception in your code in some principled way. The specific handling varies depending on the case. - (Android code style guidelines)
See alternatives here.

2.1.2 Don't catch generic exception

You should not do this:
try {
    someComplicatedIOFunction();        // may throw IOException
    someComplicatedParsingFunction();   // may throw ParsingException
    someComplicatedSecurityFunction();  // may throw SecurityException
    // phew, made it all the way
} catch (Exception e) {                 // I'll just catch all exceptions
    handleError();                      // with one generic handler!
}
See the reason why and some alternatives here

2.1.3 Don't use finalizers

We don't use finalizers. There are no guarantees as to when a finalizer will be called, or even that it will be called at all. In most cases, you can do what you need from a finalizer with good exception handling. If you absolutely need it, define a close() method (or the like) and document exactly when that method needs to be called. See InputStream for an example. In this case it is appropriate but not required to print a short log message from the finalizer, as long as it is not expected to flood the logs. - (Android code style guidelines)

2.1.4 Fully qualify imports

This is bad: import foo.*;
This is good: import foo.Bar;
See more info here

2.2 Java style rules

2.2.1 Fields definition and naming

Fields should be defined at the top of the file and they should follow the naming rules listed below.
  • Private, non-static field names start with m.
  • Private, static field names start with s.
  • Other fields start with a lower case letter.
  • Static final fields (constants) are ALL_CAPS_WITH_UNDERSCORES.
Example:
public class MyClass {
    public static final int SOME_CONSTANT = 42;
    public int publicField;
    private static MyClass sSingleton;
    int mPackagePrivate;
    private int mPrivate;
    protected int mProtected;
}

2.2.3 Treat acronyms as words

Good Bad
XmlHttpRequest XMLHTTPRequest
getCustomerId getCustomerID
String url String URL
long id long ID

2.2.4 Use spaces for indentation

Use 4 space indents for blocks:
if (x == 1) {
    x++;
}
Use 8 space indents for line wraps:
Instrument i =
        someLongExpression(that, wouldNotFit, on, one, line);

2.2.5 Use standard brace style

Braces go on the same line as the code before them.
class MyClass {
    int func() {
        if (something) {
            // ...
        } else if (somethingElse) {
            // ...
        } else {
            // ...
        }
    }
}
Braces around the statements are required unless the condition and the body fit on one line.
If the condition and the body fit on one line and that line is shorter than the max line length, then braces are not required, e.g.
if (condition) body();
This is bad:
if (condition)
    body();  // bad!

2.2.6 Annotations

2.2.6.1 Annotations practices

According to the Android code style guide, the standard practices for some of the predefined annotations in Java are:
  • @Override: The @Override annotation must be used whenever a method overrides the declaration or implementation from a super-class. For example, if you use the @inheritdocs Javadoc tag, and derive from a class (not an interface), you must also annotate that the method @Overrides the parent class's method.
  • @SuppressWarnings: The @SuppressWarnings annotation should only be used under circumstances where it is impossible to eliminate a warning. If a warning passes this "impossible to eliminate" test, the @SuppressWarnings annotation must be used, so as to ensure that all warnings reflect actual problems in the code.
More information about annotation guidelines can be found here.

2.2.6.2 Annotations style

Classes, Methods and Constructors
When annotations are applied to a class, method, or constructor, they are listed after the documentation block and should appear as one annotation per line .
/* This is the documentation block about the class */
@AnnotationA
@AnnotationB
public class MyAnnotatedClass { }
Fields
Annotations applying to fields should be listed on the same line, unless the line reaches the maximum line length.
@Nullable @Mock DataManager mDataManager;

2.2.7 Limit variable scope

The scope of local variables should be kept to a minimum (Effective Java Item 29). By doing so, you increase the readability and maintainability of your code and reduce the likelihood of error. Each variable should be declared in the innermost block that encloses all uses of the variable.
Local variables should be declared at the point they are first used. Nearly every local variable declaration should contain an initializer. If you don't yet have enough information to initialize a variable sensibly, you should postpone the declaration until you do. - (Android code style guidelines)

2.2.8 Order import statements

If you are using an IDE such as Android Studio, you don't have to worry about this because your IDE is already obeying these rules. If not, have a look below.
The ordering of import statements is:
  1. Android imports
  2. Imports from third parties (com, junit, net, org)
  3. java and javax
  4. Same project imports
To exactly match the IDE settings, the imports should be:
  • Alphabetically ordered within each grouping, with capital letters before lower case letters (e.g. Z before a).
  • There should be a blank line between each major grouping (android, com, junit, net, org, java, javax).
More info here

2.2.9 Logging guidelines

Use the logging methods provided by the Log class to print out error messages or other information that may be useful for developers to identify issues:
  • Log.v(String tag, String msg) (verbose)
  • Log.d(String tag, String msg) (debug)
  • Log.i(String tag, String msg) (information)
  • Log.w(String tag, String msg) (warning)
  • Log.e(String tag, String msg) (error)
As a general rule, we use the class name as tag and we define it as a static final field at the top of the file. For example:
public class MyClass {
    private static final String TAG = MyClass.class.getSimpleName();

    public myMethod() {
        Log.e(TAG, "My error message");
    }
}
VERBOSE and DEBUG logs must be disabled on release builds. It is also recommended to disable INFORMATION, WARNING and ERROR logs but you may want to keep them enabled if you think they may be useful to identify issues on release builds. If you decide to leave them enabled, you have to make sure that they are not leaking private information such as email addresses, user ids, etc.
To only show logs on debug builds:
if (BuildConfig.DEBUG) Log.d(TAG, "The value of x is " + x);

2.2.10 Class member ordering

There is no single correct solution for this but using a logical and consistent order will improve code learnability and readability. It is recommendable to use the following order:
  1. Constants
  2. Fields
  3. Constructors
  4. Override methods and callbacks (public or private)
  5. Public methods
  6. Private methods
  7. Inner classes or interfaces
Example:
public class MainActivity extends Activity {

    private static final String TAG = MainActivity.class.getSimpleName();

    private String mTitle;
    private TextView mTextViewTitle;

    @Override
    public void onCreate() {
        ...
    }

    public void setTitle(String title) {
     mTitle = title;
    }

    private void setUpView() {
        ...
    }

    static class AnInnerClass {

    }

}
If your class is extending an Android component such as an Activity or a Fragment, it is a good practice to order the override methods so that they match the component's lifecycle. For example, if you have an Activity that implements onCreate(), onDestroy(), onPause() and onResume(), then the correct order is:
public class MainActivity extends Activity {

 //Order matches Activity lifecycle
    @Override
    public void onCreate() {}

    @Override
    public void onResume() {}

    @Override
    public void onPause() {}

    @Override
    public void onDestroy() {}

}

2.2.11 Parameter ordering in methods

When programming for Android, it is quite common to define methods that take a Context. If you are writing a method like this, then the Context must be the first parameter.
The opposite case are callback interfaces that should always be the last parameter.
Examples:
// Context always goes first
public User loadUser(Context context, int userId);

// Callbacks always go last
public void loadUserAsync(Context context, int userId, UserCallback callback);

2.2.13 String constants, naming, and values

Many elements of the Android SDK such as SharedPreferences, Bundle, or Intent use a key-value pair approach so it's very likely that even for a small app you end up having to write a lot of String constants.
When using one of these components, you must define the keys as a static final fields and they should be prefixed as indicated below.
Element Field Name Prefix
SharedPreferences PREF_
Bundle BUNDLE_
Fragment Arguments ARGUMENT_
Intent Extra EXTRA_
Intent Action ACTION_
Note that the arguments of a Fragment - Fragment.getArguments() - are also a Bundle. However, because this is a quite common use of Bundles, we define a different prefix for them.
Example:
// Note the value of the field is the same as the name to avoid duplication issues
static final String PREF_EMAIL = "PREF_EMAIL";
static final String BUNDLE_AGE = "BUNDLE_AGE";
static final String ARGUMENT_USER_ID = "ARGUMENT_USER_ID";

// Intent-related items use full package name as value
static final String EXTRA_SURNAME = "com.myapp.extras.EXTRA_SURNAME";
static final String ACTION_OPEN_USER = "com.myapp.action.ACTION_OPEN_USER";

2.2.14 Arguments in Fragments and Activities

When data is passed into an Activity or Fragment via an Intent or a Bundle, the keys for the different values must follow the rules described in the section above.
When an Activity or Fragment expects arguments, it should provide a public static method that facilitates the creation of the relevant Intent or Fragment.
In the case of Activities the method is usually called getStartIntent():
public static Intent getStartIntent(Context context, User user) {
 Intent intent = new Intent(context, ThisActivity.class);
 intent.putParcelableExtra(EXTRA_USER, user);
 return intent;
}
For Fragments it is named newInstance() and handles the creation of the Fragment with the right arguments:
public static UserFragment newInstance(User user) {
 UserFragment fragment = new UserFragment();
 Bundle args = new Bundle();
 args.putParcelable(ARGUMENT_USER, user);
 fragment.setArguments(args)
 return fragment;
}
Note 1: These methods should go at the top of the class before onCreate().
Note 2: If we provide the methods described above, the keys for extras and arguments should be private because there is not need for them to be exposed outside the class.

2.2.15 Line length limit

Code lines should not exceed 100 characters. If the line is longer than this limit there are usually two options to reduce its length:
  • Extract a local variable or method (preferable).
  • Apply line-wrapping to divide a single line into multiple ones.
There are two exceptions where it is possible to have lines longer than 100:
  • Lines that are not possible to split, e.g. long URLs in comments.
  • package and import statements.

2.2.15.1 Line-wrapping strategies

There isn't an exact formula that explains how to line-wrap and quite often different solutions are valid. However there are a few rules that can be applied to common cases.
Break at operators
When the line is broken at an operator, the break comes before the operator. For example:
int longName = anotherVeryLongVariable + anEvenLongerOne - thisRidiculousLongOne
        + theFinalOne;
Assignment Operator Exception
An exception to the break at operators rule is the assignment operator =, where the line break should happen after the operator.
int longName =
        anotherVeryLongVariable + anEvenLongerOne - thisRidiculousLongOne + theFinalOne;
Method chain case
When multiple methods are chained in the same line - for example when using Builders - every call to a method should go in its own line, breaking the line before the .
Picasso.with(context).load("http://ribot.co.uk/images/sexyjoe.jpg").into(imageView);
Picasso.with(context)
        .load("http://ribot.co.uk/images/sexyjoe.jpg")
        .into(imageView);
Long parameters case
When a method has many parameters or its parameters are very long, we should break the line after every comma ,
loadPicture(context, "http://ribot.co.uk/images/sexyjoe.jpg", mImageViewProfilePicture, clickListener, "Title of the picture");
loadPicture(context,
        "http://ribot.co.uk/images/sexyjoe.jpg",
        mImageViewProfilePicture,
        clickListener,
        "Title of the picture");

2.2.16 RxJava chains styling

Rx chains of operators require line-wrapping. Every operator must go in a new line and the line should be broken before the .
public Observable<Location> syncLocations() {
    return mDatabaseHelper.getAllLocations()
            .concatMap(new Func1<Location, Observable<? extends Location>>() {
                @Override
                 public Observable<? extends Location> call(Location location) {
                     return mRetrofitService.getLocation(location.id);
                 }
            })
            .retry(new Func2<Integer, Throwable, Boolean>() {
                 @Override
                 public Boolean call(Integer numRetries, Throwable throwable) {
                     return throwable instanceof RetrofitError;
                 }
            });
}

2.3 XML style rules

2.3.1 Use self closing tags

When an XML element doesn't have any contents, you must use self closing tags.
This is good:
<TextView
 android:id="@+id/text_view_profile"
 android:layout_width="wrap_content"
 android:layout_height="wrap_content" />
This is bad :
<!-- Don\'t do this! -->
<TextView
    android:id="@+id/text_view_profile"
    android:layout_width="wrap_content"
    android:layout_height="wrap_content" >
</TextView>

2.3.2 Resources naming

Resource IDs and names are written in lowercase_underscore.

2.3.2.1 ID naming

IDs should be prefixed with the name of the element in lowercase underscore. For example:
Element Prefix
TextView text_
ImageView image_
Button button_
Menu menu_
Image view example:
<ImageView
    android:id="@+id/image_profile"
    android:layout_width="wrap_content"
    android:layout_height="wrap_content" />
Menu example:
<menu>
 <item
        android:id="@+id/menu_done"
        android:title="Done" />
</menu>

2.3.2.2 Strings

String names start with a prefix that identifies the section they belong to. For example registration_email_hint or registration_name_hint. If a string doesn't belong to any section, then you should follow the rules below:
Prefix Description
error_ An error message
msg_ A regular information message
title_ A title, i.e. a dialog title
action_ An action such as "Save" or "Create"

2.3.2.3 Styles and Themes

Unlike the rest of resources, style names are written in UpperCamelCase.

2.3.3 Attributes ordering

As a general rule you should try to group similar attributes together. A good way of ordering the most common attributes is:
  1. View Id
  2. Style
  3. Layout width and layout height
  4. Other layout attributes, sorted alphabetically
  5. Remaining attributes, sorted alphabetically

2.4 Tests style rules

2.4.1 Unit tests

Test classes should match the name of the class the tests are targeting, followed by Test. For example, if we create a test class that contains tests for the DatabaseHelper, we should name it DatabaseHelperTest.
Test methods are annotated with @Test and should generally start with the name of the method that is being tested, followed by a precondition and/or expected behaviour.
  • Template: @Test void methodNamePreconditionExpectedBehaviour()
  • Example: @Test void signInWithEmptyEmailFails()
Precondition and/or expected behaviour may not always be required if the test is clear enough without them.
Sometimes a class may contain a large amount of methods, that at the same time require several tests for each method. In this case, it's recommendable to split up the test class into multiple ones. For example, if the DataManager contains a lot of methods we may want to divide it into DataManagerSignInTest, DataManagerLoadUsersTest, etc. Generally you will be able to see what tests belong together because they have common test fixtures.

2.4.2 Espresso tests

Every Espresso test class usually targets an Activity, therefore the name should match the name of the targeted Activity followed by Test, e.g. SignInActivityTest
When using the Espresso API it is a common practice to place chained methods in new lines.
onView(withId(R.id.view))
        .perform(scrollTo())
        .check(matches(isDisplayed()))

Android Project Structure



Goals of the new Build System
The goals of the new build system are:
  • Make it easy to reuse code and resources
  • Make it easy to create several variants of an application, either for multi-apk distribution or for different flavors of an application
  • Make it easy to configure, extend and customize the build process
  • Good IDE integration

Why Gradle?

Gradle is an advanced build system as well as an advanced build toolkit allowing to create custom build logic through plugins. Here are some of its features that made us choose Gradle:
  • Domain Specific Language (DSL) based on Groovy, used to describe and manipulate the build logic
  • Build files are Groovy based and allow mixing of declarative elements through the DSL and using code to manipulate the DSL elements to provide custom logic.
  • Built-in dependency management through Maven and/or Ivy.
  • Very flexible. Allows using best practices but doesn’t force its own way of doing things.
  • Plugins can expose their own DSL and their own API for build files to use.
  • Good Tooling API allowing IDE integration

Requirements

  • Gradle 2.2
  • SDK with Build Tools 19.0.0. Some features may require a more recent version.


Basic Project Setup

A Gradle project describes its build in a file called build.gradle located in the root folder of the project. (See Gradle User Guide for an overview of the build system itself.)

Simple build files

The most simple Android project has the following build.gradle:



buildscript {
    repositories {
        jcenter()

    }


    dependencies {
        classpath 'com.android.tools.build:gradle:1.3.1'
    }
}


apply plugin: 'com.android.application'


android {
    compileSdkVersion 23
    buildToolsVersion "23.1.0"
}

There are 3 main areas to this Android build file:

buildscript { ... } configures the code driving the build. In this case, this declares that it uses the jCenter repository, and that there is a classpath dependency on a Maven artifact. This artifact is the library that contains the Android plugin for Gradle in version 1.3.1. Note: This only affects the code running the build, not the project. The project itself needs to declare its own repositories and dependencies. This will be covered later.

Then, the com.android.application plugin is applied. This is the plugin used for building Android applications.

Finally, android { ... } configures all the parameters for the android build. This is the entry point for the Android DSL. By default, only the compilation target, and the version of the build-tools are needed. This is done with the compileSdkVersion and buildtoolsVersion properties.
The compilation target is the same as the target property in the project.properties file of the old build system. This new property can either be assigned a int (the api level) or a string with the same value as the previous target property.

Important: You should only apply the com.android.application plugin. Applying the java plugin as well will result in a build error.

Note: You will also need a local.properties file to set the location of the SDK in the same way that the existing SDK requires, using the sdk.dir property.
Alternatively, you can set an environment variable called ANDROID_HOME. There is no differences between the two methods, you can use the one you prefer. Example local.properties file:

sdk.dir=/path/to/Android/Sdk

Project Structure

The basic build files above expects a default folder structure. Gradle follows the concept of convention over configuration, providing sensible default option values when possible. The basic project starts with two components called “source sets”, one for the main source code and one for the test code. These live respectively in:
  • src/main/
  • src/androidTest/
Inside each of these directories there are subdirectories for each source component. For both the Java and Android plugin, the location of the Java source code and the Java resources are:
  • java/
  • resources/
For the Android plugin, extra files and folders specific to Android:
  • AndroidManifest.xml
  • res/
  • assets/
  • aidl/
  • rs/
  • jni/
  • jniLibs/
This means that *.java files for the main source set are located in src/main/java and the main manifest is src/main/AndroidManifest.xml

Note: src/androidTest/AndroidManifest.xml is not needed as it is created automatically.

Configuring the Structure

When the default project structure isn’t adequate, it is possible to configure it. See this page in gradle documentation for information how this can be done for pure-java projects.

The Android plugin uses a similar syntax, but because it uses its own sourceSets, this is done within the android block. Here’s an example, using the old project structure (used in Eclipse) for the main code and remapping the androidTest sourceSet to the tests folder:

android {
    sourceSets {
        main {
            manifest.srcFile 'AndroidManifest.xml'
            java.srcDirs = ['src']
            resources.srcDirs = ['src']
            aidl.srcDirs = ['src']
            renderscript.srcDirs = ['src']
            res.srcDirs = ['res']
            assets.srcDirs = ['assets']
        }

        androidTest.setRoot('tests')
    }
}

Note: because the old structure put all source files (Java, AIDL and RenderScript) in the same folder, we need to remap all those new components of the sourceSet to the same src folder.
Note: setRoot() moves the whole sourceSet (and its sub folders) to a new folder. This moves src/androidTest/* to tests/* This is Android specific and will not work on Java sourceSets.

Build Tasks

General Tasks

Applying a plugin to the build file automatically creates a set of build tasks to run. Both the Java plugin and the Android plugin do this. The convention for tasks is the following:
  • assemble
    The task to assemble the output(s) of the project.
  • check
    The task to run all the checks.
  • build
    This task does both assemble and check.
  • clean
    This task cleans the output of the project.
The tasks assemblecheck and build don’t actually do anything. They are "anchor" tasks for the plugins to add dependent tasks that actually do the work.

This allows you to always call the same task(s) no matter what the type of project is, or what plugins are applied. For instance, applying the findbugs plugin will create a new task and make check depend on it, making it be called whenever the check task is called.

From the command line you can get the high level task running the following command:
gradle tasks

For a full list and seeing dependencies between the tasks run:
gradle tasks --all

Note: Gradle automatically monitor the declared inputs and outputs of a task.

Running the build task twice without making changes to your project, will make Gradle report all tasks as UP-TO-DATE, meaning no work was required. This allows tasks to properly depend on each other without requiring unnecessary build operations.

Java project tasks

Here are the two most important tasks created by the java plugin, dependencies of the main anchor tasks:
  • assemble
    • jar
      This task creates the output.
  • check
    • test
      This task runs the tests.
The jar task itself will depend directly and indirectly on other tasks: classes for instance will compile the Java code. The tests are compiled with testClasses, but it is rarely useful to call this as test depends on it (as well as classes).

In general, you will probably only ever call assemble or check, and ignore the other tasks. You can see the full set of tasks and their descriptions for the Java plugin here.

Android tasks

The Android plugin uses the same convention to stay compatible with other plugins, and adds an additional anchor task:
  • assemble
    The task to assemble the output(s) of the project.
  • check
    The task to run all the checks.
  • connectedCheck
    Runs checks that requires a connected device or emulator. they will run on all connected devices in parallel.
  • deviceCheck
    Runs checks using APIs to connect to remote devices. This is used on CI servers.
  • build
    This task does both assemble and check
  • clean
    This task cleans the output of the project.
The new anchor tasks are necessary in order to be able to run regular checks without needing a connected device. Note that build does not depend on deviceCheck, or connectedCheck.

An Android project has at least two outputs: a debug APK and a release APK. Each of these has its own anchor task to facilitate building them separately:
  • assemble
    • assembleDebug
    • assembleRelease
They both depend on other tasks that execute the multiple steps needed to build an APK. The assemble task depends on both, so calling it will build both APKs.

Tip: Gradle support camel case shortcuts for task names on the command line. For instance:
gradle aR
is the same as typing
gradle assembleRelease
as long as no other task matches ‘aR’

The check anchor tasks have their own dependencies:
  • check
    • lint
  • connectedCheck
    • connectedAndroidTest
  • deviceCheck
    • This depends on tasks created when other plugins implement test extension points.
Finally, the plugin creates tasks for installing and uninstalling all build types (debug, release, test), as long as they can be installed (which requires signing), e.g.
  • installDebug
  • installRelease
  • uninstallAll
    • uninstallDebug
    • uninstallRelease
    • uninstallDebugAndroidTest

Basic Build Customization

The Android plugin provides a broad DSL to customize most things directly from the build system.

Manifest entries

Through the DSL it is possible to configure the most important manifest entries, like these:
  • minSdkVersion
  • targetSdkVersion
  • versionCode
  • versionName
  • applicationId (the effective packageName -- see ApplicationId versus PackageName for more information)
  • testApplicationId (used by the test APK)
  • testInstrumentationRunner
Example:
android {
    compileSdkVersion 23
    buildToolsVersion "23.0.1"


    defaultConfig {
        versionCode 12

        versionName "2.0"
        minSdkVersion 16
        targetSdkVersion 23
    }
}

Please see the Android Plugin DSL Reference for a complete list of build properties that can be configured and their default values.

The power of putting these manifest properties in the build file is that the values can be chosen dynamically. For instance, one could be reading the version name from a file or using other custom logic:


def computeVersionName() {
    ...
}


android {
    compileSdkVersion 23
    buildToolsVersion "23.0.1"


    defaultConfig {
        versionCode 12
        versionName computeVersionName()

        minSdkVersion 16
        targetSdkVersion 23
    }
}


Note: Do not use function names that could conflict with existing getters in the given scope. For instance defaultConfig { ...} calling getVersionName() will automatically use the getter of defaultConfig.getVersionName() instead of the custom method.

Build Types

By default, the Android plugin automatically sets up the project to build both a debug and a release version of the application. These differ mostly around the ability to debug the application on a secure (non dev) devices, and details of how the APK is signed. The debug version is signed with a key/certificate that is created automatically with a known name/password (to prevent required prompt during the build). The release is not signed during the build, this needs to happen after.

This configuration is done through an object called a BuildType. By default, 2 instances are created, a debug and a release one. The Android plugin allows customizing those two instances as well as creating other Build Types. This is done with the buildTypes DSL container:

android {
    buildTypes {
        debug {
            applicationIdSuffix ".debug"
        }


        jnidebug {
            initWith(buildTypes.debug)
            applicationIdSuffix ".jnidebug"
            jniDebuggable true
        }
    }
}

The above snippet achieves the following:
  • Configures the default debug Build Type:
    • set its package to be <app appliationId>.debug to be able to install both debug and release apk on the same device
  • Creates a new BuildType called jnidebug and configure it to be a copy of the debug build type.
  • Keep configuring the jnidebug, by enabling debug build of the JNI component, and add a different package suffix.
Creating new Build Types is as easy as using a new element under the buildTypes container, either to call initWith() or to configure it with a closure. See the DSL Reference for a list of all properties that can be configured on a build type.

In addition to modifying build properties, Build Types can be used to add specific code and resources. For each Build Type, a new matching sourceSet is created, with a default location of src/<buildtypename>/, e.g. src/debug/java directory can be used to add sources that will only be compiled for the debug APK. This means the Build Type names cannot be main or androidTest (this is enforced by the plugin), and that they have to be unique.

Like any other source sets, the location of the build type source set can be relocated:

android {
    sourceSets.jnidebug.setRoot('foo/jnidebug')
}

Additionally, for each Build Type, a new assemble<BuildTypeName> task is created, e.g. assembleDebugThe assembleDebug and assembleRelease tasks have already been mentioned, and this is where they come from. When the debug and release Build Types are pre-created, their tasks are automatically created as well. According to this rule, build.gradle snippet above will also generate an assembleJnidebug task, and assemble would be made to depend on it the same way it depends on the assembleDebug and assembleRelease tasks.

Tip: remember that you can type gradle aJ to run the assembleJnidebug task.

Possible use case:
  • Permissions in debug mode only, but not in release mode
  • Custom implementation for debugging
  • Different resources for debug mode (for instance when a resource value is tied to the signing certificate).
The code/resources of the BuildType are used in the following way:
  • The manifest is merged into the app manifest
  • The code acts as just another source folder
  • The resources are overlayed over the main resources, replacing existing values.

Signing Configurations

Signing an application requires the following (See Signing Your Application for details about signing an APK):
  • A keystore
  • A keystore password
  • A key alias name
  • A key password
  • The store type
The location, as well as the key name, both passwords and store type form together a Signing Configuration. By default, there is a debug configuration that is setup to use a debug keystore, with a known password and a default key with a known password.
The debug keystore is located in $HOME/.android/debug.keystore, and is created if not present. The debug Build Type is set to use this debug SigningConfig automatically.

It is possible to create other configurations or customize the default built-in one. This is done through the signingConfigs DSL container:
android {
    signingConfigs {
        debug {
            storeFile file("debug.keystore")
        }


        myConfig {
            storeFile file("other.keystore")
            storePassword "android"
            keyAlias "androiddebugkey"
            keyPassword "android"
        }
    }


    buildTypes {
        foo {
            signingConfig signingConfigs.myConfig
        }
    }
}

The above snippet changes the location of the debug keystore to be at the root of the project. This automatically impacts any Build Types that are set to using it, in this case the debug Build Type. It also creates a new Signing Config and a new Build Type that uses the new configuration.

Note: Only debug keystores located in the default location will be automatically created. Changing the location of the debug keystore will not create it on-demand. Creating a SigningConfig with a different name that uses the default debug keystore location will create it automatically. In other words, it’s tied to the location of the keystore, not the name of the configuration.

Note: Location of keystores are usually relative to the root of the project, but could be absolute paths, thought it is not recommended (except for the debug one since it is automatically created).

Note:  If you are checking these files into version control, you may not want the password in the file. The following Stack Overflow post shows ways to read the values from the console, or from environment variables.


Dependencies, Android Libraries and Multi-project setup

Gradle projects can have dependencies on other components. These components can be external binary packages, or other Gradle projects.

Dependencies on binary packages

Local packages

To configure a dependency on an external library jar, you need to add a dependency on the compile configuration. The snippet below adds a dependency on all jars inside the libs directory.

dependencies {
    compile fileTree(dir: 'libs', include: ['*.jar'])
}


android {
    ...
}

Note: the dependencies DSL element is part of the standard Gradle API and does not belong inside the android element.

The compile configuration is used to compile the main application. Everything in it is added to the compilation classpath and also packaged in the final APK. There are other possible configurations to add dependencies to:
  • compile: main application
  • androidTestCompile: test application
  • debugCompile: debug Build Type
  • releaseCompile: release Build Type.
Because it’s not possible to build an APK that does not have an associated Build Type, the APK is always configured with two (or more) configurations: compile and <buildtype>CompileCreating a new Build Type automatically creates a new configuration based on its name. This can be useful if the debug version needs to use a custom library (to report crashes for instance), while the release doesn’t, or if they rely on different versions of the same library (see Gradle documentation on details of how version conflicts are handled).

Remote artifacts

Gradle supports pulling artifacts from Maven and Ivy repositories. First the repository must be added to the list, and then the dependency must be declared in a way that Maven or Ivy declare their artifacts.

repositories {
     jcenter()
}


dependencies {
    compile 'com.google.guava:guava:18.0'
}


android {
    ...
}

Note: jcenter() is a shortcut to specifying the URL of the repository. Gradle supports both remote and local repositories.
Note: Gradle will follow all dependencies transitively. This means that if a dependency has dependencies of its own, those are pulled in as well.

For more information about setting up dependencies, read the Gradle user guide here, and DSL documentation here.

Multi project setup

Gradle projects can also depend on other gradle projects by using a multi-project setup. A multi-project setup usually works by having all the projects as sub folders of a given root project. For instance, given to following structure:
MyProject/
 + app/
 + libraries/
    + lib1/
    + lib2/

We can identify 3 projects. Gradle will reference them with the following name:
:app
:libraries:lib1
:libraries:lib2

Each projects will have its own build.gradle declaring how it gets built. Additionally, there will be a file called settings.gradle at the root declaring the projects. This gives the following structure:
MyProject/
 | settings.gradle
 + app/
    | build.gradle
 + libraries/
    + lib1/
       | build.gradle
    + lib2/
       | build.gradle

The content of settings.gradle is very simple. It defines which folder is actually a Gradle project:

include ':app', ':libraries:lib1', ':libraries:lib2'

The :app project is likely to depend on the libraries, and this is done by declaring the following dependencies:

dependencies {
     compile project(':libraries:lib1')
}

More general information about multi-project setup here.

Library projects

In the above multi-project setup, :libraries:lib1 and :libraries:lib2 can be Java projects, and the :app Android project will use their jar output. However, if you want to share code that accesses Android APIs or uses Android-style resources, these libraries cannot be regular Java project, they have to be Android Library Projects.

Creating a Library Project

A Library project is very similar to a regular Android project with a few differences. Since building libraries is different than building applications, a different plugin is used. Internally both plugins share most of the same code and they are both provided by the same com.android.tools.build.gradle jar.



buildscript {
    repositories {
        jcenter()
    }


    dependencies {
        classpath 'com.android.tools.build:gradle:1.3.1'
    }
}


apply plugin: 'com.android.library'


android {
    compileSdkVersion 23
    buildToolsVersion "23.0.1"
}


This creates a library project that uses API 23 to compile. SourceSets, build types and dependencies are handled the same as they are in an application project and can be customized the same way.

Differences between a Project and a Library Project

A Library project's main output is an .aar package (which stands for Android archive). It is a combination of compile code (as a jar file and/or native .so files) and resources (manifest, res, assets). A library project can also generate a test apk to test the library independently from an application. The same anchor tasks are used for this (assembleDebug, assembleRelease) so there’s no difference in commands to build such a project. For the rest, libraries behave the same as application projects. They have build types and product flavors (see below), and can potentially generate more than one version of the aar. Note that most of the configuration of the Build Type do not apply to library projects. However you can use the custom sourceSet to change the content of the library depending on whether it’s used by a project or being tested.

Referencing a Library

Referencing a library is done the same way any other project is referenced:

dependencies {
    compile project(':libraries:lib1')
    compile project(':libraries:lib2')
}

Note: if you have more than one library, then the order will be important. This is similar to the old build system where the order of the dependencies in the project.properties file was important.

Library Publication

By default a library only publishes its release variant. This variant will be used by all projects referencing the library, no matter which variant they build themselves. This is a temporary limitation due to Gradle limitations that we are working towards removing. You can control which variant gets published:

android {
    defaultPublishConfig "debug"
}

Note that this publishing configuration name references the full variant name. Release and debug are only applicable when there are no flavors. If you wanted to change the default published variant while using flavors, you would write:

android {
    defaultPublishConfig "flavor1Debug"
}

It is also possible to publish all variants of a library. We are planning to allow this while using a normal project-to-project dependency (like shown above), but this is not possible right now due to limitations in Gradle (we are working toward fixing those as well).
Publishing of all variants are not enabled by default. The snippet below enables this feature:

android {
    publishNonDefault true
}

It is important to realize that publishing multiple variants means publishing multiple aar files, instead of a single aar containing multiple variants. Each aar packaging contains a single variant. Publishing a variant means making this aar available as an output artifact of the Gradle project. This can then be used either when publishing to a maven repository, or when another project creates a dependency on the library project.

Gradle has a concept of default" artifact. This is the one that is used when writing:

dependencies {
    compile project(':libraries:lib2')
}

To create a dependency on another published artifact, you need to specify which one to use:

dependencies {
    flavor1Compile project(path: ':lib1', configuration: 'flavor1Release')
    flavor2Compile project(path: ':lib1', configuration: 'flavor2Release')
}

Important:
Note that the published configuration is a full variant, including the build type, and needs to be referenced as such. 
Important: When enabling publishing of non default, the Maven publishing plugin will publish these additional variants as extra packages (with classifier). This means that this is not really compatible with publishing to a maven repository. You should either publish a single variant to a repository OR enable all config publishing for inter-project dependencies.


Testing

Building a test application is integrated into the application project. There is no need for a separate test project anymore. 

Unit testing

For the unit testing support added in 1.1, please see this separate page. The rest of this section describes "instrumentation tests" that can run on a real device (or an emulator) and require a separate, testing APK to be built.

Basics and Configuration

As mentioned previously, next to the main sourceSet is the androidTest sourceSet, located by default in src/androidTest/. Using this sourceSet a test APK gets built, which can be deployed to a device to test the application using the Android testing framework. This can contain android unit tests, instrumentation tests, and uiautomator tests. The <instrumentation> node of the manifest for the test app is generated but you can create a src/androidTest/AndroidManifest.xml file to add other components to the test manifest.

There are a few values that can be configured for the test app, in order to configure the <instrumentation> node (see the DSL reference for details)
  • testApplicationId
  • testInstrumentationRunner
  • testHandleProfiling
  • testFunctionalTest
As seen previously, those are configured in the defaultConfig object:
android {
    defaultConfig {
        testApplicationId "com.test.foo"
        testInstrumentationRunner "android.test.InstrumentationTestRunner"
        testHandleProfiling true
        testFunctionalTest true
    }
}

The value of the targetPackage attribute of the instrumentation node in the test application manifest is automatically filled with the package name of the tested app, even if it is customized through the defaultConfig and/or the Build Type objects. This is one of the reason this part of the manifest is generated automatically.

Additionally, the androidTest source set can be configured to have its own dependencies. By default, the application and its own dependencies are added to the test app classpath, but this can be extended with the snippet below:
dependencies {
    androidTestCompile 'com.google.guava:guava:11.0.2'
}

The test app is built by the assembleAndroidTest task. It is not a dependency of the main assemble task, and is instead called automatically when the tests are set to run.

Currently only one Build Type is tested. By default it is the debug Build Type, but this can be reconfigured with:
android {
    ...
    testBuildType "staging"
}

Resolving conflicts between main and test APK

When instrumentation tests are run, both the main APK and test APK share the same classpath. Gradle build will fail if the main APK and the test APK use the same library (e.g. Guava) but in different versions. If gradle didn't catch that, your app could behave differently during tests and during normal run (including crashing in one of the cases).

To make the build succeed, just make sure both APKs use the same version. If the error is about an indirect dependency (a library you didn't mention in your build.gradle), just add a dependency for the newer version to the configuration ("compile" or "androidTestCompile") that needs it. You can also use Gradle's resolution strategy mechanismYou can inspect the dependency tree by running ./gradlew :app:dependencies and ./gradlew :app:androidDependencies.

Running tests

As mentioned previously, checks requiring a connected device are launched with the anchor task called connectedCheckThis depends on the task connectedDebugAndroidTest and therefore will run it. This task does the following:
  • Ensure the app and the test app are built (depending on assembleDebug and assembleDebugAndroidTest).
  • Install both apps.
  • Run the tests.
  • Uninstall both apps.
If more than one device is connected, all tests are run in parallel on all connected devices. If one of the test fails, on any device, the build will fail.

Testing Android Libraries

Testing Android Library project is done exactly the same way as application projects. The only difference is that the whole library (and its dependencies) is automatically added as a Library dependency to the test app. The result is that the test APK includes not only its own code, but also the library itself and all its dependencies. The manifest of the Library is merged into the manifest of the test app (as is the case for any project referencing this Library). The androidTest task is changed to only install (and uninstall) the test APK (since there are no other APK to install.) and everything else is identical.

Test reports

When running unit tests, Gradle outputs an HTML report to easily look at the results. The Android plugins build on this and extends the HTML report to aggregate the results from all connected devices. All test results are stored as XML files under build/reports/androidTests/ (this is similar to regular jUnit results that are stored under build/reports/tests). This can be configured with:

android {
    ...

    testOptions {
        resultsDir = "${project.buildDir}/foo/results"
    }
}

The value of android.testOptions.resultsDir is evaluated with Project.file(String)

Multi-projects reports

In a multi project setup with application(s) and library(ies) projects, when running all tests at the same time, it might be useful to generate a single reports for all tests.

To do this, a different plugin is available in the same artifact. It can be applied with:
buildscript {
    repositories {
        jcenter()
    }


    dependencies {
        classpath 'com.android.tools.build:gradle:0.5.6'
    }
}


apply plugin: 'android-reporting'

This should be applied to the root project, ie in build.gradle next to settings.gradle

Then from the root folder, the following command line will run all the tests and aggregate the reports:
gradle deviceCheck mergeAndroidReports --continue

Note: the --continue option ensure that all tests, from all sub-projects will be run even if one of them fails. Without it the first failing test will interrupt the run and not all projects may have their tests run.

Lint support

You can run lint for a specific variant (see below), e.g. ./gradlew lintRelease, or for all variants (./gradlew lint), in which case it produces a report which describes which specific variants a given issue applies to. You can configure lint by adding a lintOptions section like following. You typically only specify a few of these, see the DSL reference for all available options.

android {
    lintOptions {
        // turn off checking the given issue id's
        disable 'TypographyFractions','TypographyQuotes'

        // turn on the given issue id's
        enable 'RtlHardcoded','RtlCompat', 'RtlEnabled'

        // check *only* the given issue id's
        check 'NewApi', 'InlinedApi'
    }
}


Build Variants

One goal of the new build system is to enable creating different versions of the same application.

There are two main use cases:
  1. Different versions of the same application
    For instance, a free/demo version vs the “pro” paid application.
  2. Same application packaged differently for multi-apk in Google Play Store.
    See http://developer.android.com/google/play/publishing/multiple-apks.html for more information.
  3. A combination of 1. and 2. 
The goal was to be able to generate these different APKs from the same project, as opposed to using a single Library Projects and 2+ Application Projects.

Product flavors

A product flavor defines a customized version of the application build by the project. A single project can have different flavors which change the generated application.

This new concept is designed to help when the differences are very minimum. If the answer to “Is this the same application?” is yes, then this is probably the way to go over Library Projects.

Product flavors are declared using a productFlavors DSL container:

android {
    ....


    productFlavors {
        flavor1 {
            ...
        }


        flavor2 {
            ...
        }
    }
}

This creates two flavors, called flavor1 and flavor2

Note: The name of the flavors cannot collide with existing Build Type names, or with the androidTest and test source sets.

Build Type + Product Flavor = Build Variant

As we have seen before, each Build Type generates a new APK. Product Flavors do the same: the output of the project becomes all possible combinations of Build Types and, if applicable, Product Flavors. Each (Build Type, Product Flavor) combination is called a Build Variant. For instance, with the default debug and release Build Types, the above example generates four Build Variants:
  • Flavor1 - debug
  • Flavor1 - release
  • Flavor2 - debug
  • Flavor2 - release
Projects with no flavors still have Build Variants, but the single default flavor/config is used, nameless, making the list of variants similar to the list of Build Types.

Product Flavor Configuration

Each flavors is configured with a closure:

android {
    ...


    defaultConfig {
        minSdkVersion 8
        versionCode 10
    }


    productFlavors {
        flavor1 {
            applicationId "com.example.flavor1"
            versionCode 20
         }


         flavor2 {
             applicationId "com.example.flavor2"
             minSdkVersion 14
         }
    }
}

Note that the android.productFlavors.* objects are of type ProductFlavor which is the same type as the android.defaultConfig object. This means they share the same properties.

defaultConfig provides the base configuration for all flavors and each flavor can override any value. In the example above, the configurations end up being:

  • flavor1
    • applicationId: com.example.flavor1
    • minSdkVersion: 8
    • versionCode: 20
  • flavor2
    • applicationId: com.example.flavor2
    • minSdkVersion: 14
    • versionCode: 10
Usually, the Build Type configuration is an overlay over the other configuration. For instance, the Build Type's applicationIdSuffix is appended to the Product Flavor's applicationId. There are cases where a setting is settable on both the Build Type and the Product Flavor. In this case, it’s is on a case by case basis. For instance, signingConfig is one of these properties. This enables either having all release packages share the same SigningConfig, by setting android.buildTypes.release.signingConfig, or have each release package use their own SigningConfig by setting each android.productFlavors.*.signingConfig objects separately.

Sourcesets and Dependencies

Similar to Build Types, Product Flavors also contribute code and resources through their own sourceSets. The above example creates four sourceSets:
  • android.sourceSets.flavor1
    Location src/flavor1/
  • android.sourceSets.flavor2
    Location src/flavor2/
  • android.sourceSets.androidTestFlavor1
    Location src/androidTestFlavor1/
  • android.sourceSets.androidTestFlavor2
    Location src/androidTestFlavor2/
Those sourceSets are used to build the APK, alongside android.sourceSets.main and the Build Type sourceSet. The following rules are used when dealing with all the sourcesets used to build a single APK:
  • All source code (src/*/java) are used together as multiple folders generating a single output.
  • Manifests are all merged together into a single manifest. This allows Product Flavors to have different components and/or permissions, similarly to Build Types.
  • All resources (Android res and assets) are used using overlay priority where the Build Type overrides the Product Flavor, which overrides the main sourceSet.
  • Each Build Variant generates its own R class (or other generated source code) from the resources. Nothing is shared between variants.

Finally, like Build Types, Product Flavors can have their own dependencies. For instance, if the flavors are used to generate a ads-based app and a paid app, one of the flavors could have a dependency on an Ads SDK, while the other does not.

dependencies {
    flavor1Compile "..."
}

In this particular case, the file src/flavor1/AndroidManifest.xml would probably need to include the internet permission.

Additional sourcesets are also created for each variants:
  • android.sourceSets.flavor1Debug
    Location src/flavor1Debug/
  • android.sourceSets.flavor1Release
    Location src/flavor1Release/
  • android.sourceSets.flavor2Debug
    Location src/flavor2Debug/
  • android.sourceSets.flavor2Release
    Location src/flavor2Release/
These have higher priority than the build type sourcesets, and allow customization at the variant level.

Building and Tasks

We previously saw that each Build Type creates its own assemble<name> task, but that Build Variants are a combination of Build Type and Product Flavor.

When Product Flavors are used, more assemble-type tasks are created. These are:
  1. assemble<Variant Name>
  2. assemble<Build Type Name>
  3. assemble<Product Flavor Name>
#1 allows directly building a single variant. For instance assembleFlavor1Debug.

#2 allows building all APKs for a given Build Type. For instance assembleDebug will build both Flavor1Debug and Flavor2Debug variants.

#3 allows building all APKs for a given flavor. For instance assembleFlavor1 will build both Flavor1Debug and Flavor1Release variants.

The task assemble will build all possible variants.

Multi-flavor variants

In some case, one may want to create several versions of the same apps based on more than one criteria.
For instance, multi-apk support in Google Play supports 4 different filters. Creating different APKs split on each filter requires being able to use more than one dimension of Product Flavors.

Consider the example of a game that has a demo and a paid version and wants to use the ABI filter in the multi-apk support. With 3 ABIs and two versions of the application, 6 APKs needs to be generated (not counting the variants introduced by the different Build Types).
However, the code of the paid version is the same for all three ABIs, so creating simply 6 flavors is not the way to go.
Instead, there are two dimensions of flavors, and variants should automatically build all possible combinations.

This feature is implemented using Flavor Dimensions. Flavors are assigned to a specific dimension:
android {
    ...


    flavorDimensions "abi", "version"


    productFlavors {
        freeapp {
            dimension "version"
            ...
        }

        paidapp {
            dimension "version"
            ...
        }


        arm {
            dimension "abi"
            ...
        }

        mips {
            dimension "abi"
            ...
        }

        x86 {
            dimension "abi"
            ...
        }
    }
}

The android.flavorDimensions array defines the possible dimensions, as well as the order. Each defined Product Flavor is assigned to a dimension.

From the following dimensioned Product Flavors [freeapp, paidapp] and [x86, arm, mips] and the [debug, release] Build Types, the following build variants will be created:
  • x86-freeapp-debug
  • x86-freeapp-release
  • arm-freeapp-debug
  • arm-freeapp-release
  • mips-freeapp-debug
  • mips-freeapp-release
  • x86-paidapp-debug
  • x86-paidapp-release
  • arm-paidapp-debug
  • arm-paidapp-release
  • mips-paidapp-debug
  • mips-paidapp-release
The order of the dimension as defined by android.flavorDimensions is very important.

Each variant is configured by several Product Flavor objects:
  • android.defaultConfig
  • One from the abi dimension
  • One from the version dimension
The order of the dimension drives which flavor override the other, which is important for resources when a value in a flavor replaces a value defined in a lower priority flavor.
The flavor dimension is defined with higher priority first. So in this case:
abi > version > defaultConfig

Multi-flavors projects also have additional sourcesets, similar to the variant sourcesets but without the build type:

  • android.sourceSets.x86Freeapp
    Location src/x86Freeapp/
  • android.sourceSets.armPaidapp
    Location src/armPaidapp/
  • etc...
These allow customization at the flavor-combination level. They have higher priority than the basic flavor sourcesets, but lower priority than the build type sourcesets.

Testing

Testing multi-flavors project is very similar to simpler projects.

The androidTest sourceset is used for common tests across all flavors, while each flavor can also have its own tests.

As mentioned above, sourceSets to test each flavor are created:
  • android.sourceSets.androidTestFlavor1
    Location src/androidTestFlavor1/
  • android.sourceSets.androidTestFlavor2
    Location src/androidTestFlavor2/

Similarly, those can have their own dependencies:
dependencies {
    androidTestFlavor1Compile "..."
}

Running the tests can be done through the main deviceCheck anchor task, or the main androidTest tasks which acts as an anchor task when flavors are used.

Each flavor has its own task to run tests: androidTest<VariantName>. For instance:
  • androidTestFlavor1Debug
  • androidTestFlavor2Debug
Similarly, test APK building tasks and install/uninstall tasks are per variant:
  • assembleFlavor1Test
  • installFlavor1Debug
  • installFlavor1Test
  • uninstallFlavor1Debug
  • ...
Finally the HTML report generation supports aggregation by flavor.
The location of the test results and reports is as follows, first for the per flavor version, and then for the aggregated one:
  • build/androidTest-results/flavors/<FlavorName>
  • build/androidTest-results/all/
  • build/reports/androidTests/flavors<FlavorName>
  • build/reports/androidTests/all/
Customizing either path, will only change the root folder and still create sub folders per-flavor and aggregated results/reports.

BuildConfig

At compilation time, Android Studio generates a class called BuildConfig that contains constant values used when building a particular variant. You can inspect the values of these constants to change behavior in different variants, e.g.:

private void javaCode() {
    if (BuildConfig.FLAVOR.equals("paidapp")) {
        doIt();
    else {
        showOnlyInPaidAppDialog();
    }
}

Here are the values that BuildConfig contains:
  • boolean DEBUG – if the build is debuggable.
  • int VERSION_CODE
  • String VERSION_NAME
  • String APPLICATION_ID
  • String BUILD_TYPE – name of the build type, e.g. "release"
  • String FLAVOR – name of the flavor, e.g. "paidapp"
If the project uses flavor dimensions, additional values are generated. Using the example above, here's an example BuildConfig:
  • String FLAVOR = "armFreeapp"
  • String FLAVOR_abi = "arm"
  • String FLAVOR_version = "freeapp"

Filtering Variants

When you add dimensions and flavors, you can end up with variants that don't make sense. For example you may define a flavor that uses your Web API and a flavor that uses hard-coded fake data, for faster testing. The second flavor is only useful for development, but not in release builds. You can remove this variant using the variantFilter closure, like this:

android {
    productFlavors {
        realData
        fakeData
    }

    variantFilter { variant ->
        def names = variant.flavors*.name

        if (names.contains("fakeData") && variant.buildType.name == "release") {
            variant.ignore = true
        }
    }
}

With the configuration above, your project will have only three variants:
  • realDataDebug
  • realDataRelease
  • fakeDataDebug
See the DSL reference for all properties of variant that you can check.


Advanced Build Customization

Running ProGuard

The ProGuard plugin is applied automatically by the Android plugin, and the tasks are created automatically if the Build Type is configured to run ProGuard through the minifyEnabled property.

android {
    buildTypes {
        release {
            minifyEnabled true
            proguardFile getDefaultProguardFile('proguard-android.txt')
        }
    }

    productFlavors {
        flavor1 {
        }
        flavor2 {
            proguardFile 'some-other-rules.txt'
        }
    }
}

Variants use all the rules files declared in their build type, and product flavors.

There are 2 default rules files
  • proguard-android.txt
  • proguard-android-optimize.txt
They are located in the SDK. Using getDefaultProguardFile() will return the full path to the files. They are identical except for enabling optimizations.

Shrinking Resources

You can also remove unused resources, automatically, at build time. For more information, see the Resource Shrinking document.

Manipulating tasks
Basic Java projects have a finite set of tasks that all work together to create an output.
The classes task is the one that compile the Java source code.
It’s easy to access from build.gradle by simply using classes in a script. This is a shortcut for project.tasks.classes.

In Android projects, this is a bit more complicated because there could be a large number of the same task and their name is generated based on the Build Types and Product Flavors.

In order to fix this, the android object has two properties:
  • applicationVariants (only for the app plugin)
  • libraryVariants (only for the library plugin)
  • testVariants (for both plugins)
All three return a DomainObjectCollection of ApplicationVariant, LibraryVariant, and TestVariant objects respectively.

Note that accessing any of these collections will trigger the creations of all the tasks. This means no (re)configuration should take place after accessing the collections.

The DomainObjectCollection gives access to all the objects directly, or through filters which can be convenient.



android.applicationVariants.all { variant ->
   ....
}


All three variant classes share the following properties:

 Property Name  Property Type  Description
 name  String  The name of the variant. Guaranteed to be unique.
 description  String  Human readable description of the variant.
 dirName  String  subfolder name for the variant. Guaranteed to be unique. Maybe more than one folder, ie “debug/flavor1”
 baseName  String  Base name of the output(s) of the variant. Guaranteed to be unique.
 outputFile  File  The output of the variant. This is a read/write property
 processManifest  ProcessManifest  The task that processes the manifest.
 aidlCompile  AidlCompile  The task that compiles the AIDL files.
 renderscriptCompile  RenderscriptCompile  The task that compiles the Renderscript files.
 mergeResources  MergeResources  The task that merges the resources.
 mergeAssets  MergeAssets  The task that merges the assets.
 processResources  ProcessAndroidResources  The task that processes and compile the Resources.
 generateBuildConfig  GenerateBuildConfig  The task that generates the BuildConfig class.
 javaCompile  JavaCompile  The task that compiles the Java code.
 processJavaResources  Copy  The task that process the Java resources.
 assemble  DefaultTask  The assemble anchor task for this variant.

The ApplicationVariant class adds the following:

 Property Name  Property Type  Description
 buildType  BuildType  The BuildType of the variant.
 productFlavors  List<ProductFlavor>  The ProductFlavors of the variant. Always non Null but could be empty.
 mergedFlavor  ProductFlavor  The merging of android.defaultConfig and variant.productFlavors
 signingConfig  SigningConfig  The SigningConfig object used by this variant
 isSigningReady  boolean  true if the variant has all the information needed to be signed.
 testVariant  BuildVariant  The TestVariant that will test this variant.
 dex  Dex  The task that dex the code. Can be null if the variant is a library.
 packageApplication  PackageApplication  The task that makes the final APK. Can be null if the variant is a library.
 zipAlign  ZipAlign  The task that zipaligns the apk. Can be null if the variant is a library or if the APK cannot be signed.
 install  DefaultTask  The installation task. Can be null.
 uninstall  DefaultTask  The uninstallation task.

The LibraryVariant class adds the following:

 Property Name  Property Type  Description
 buildType  BuildType  The BuildType of the variant.
 mergedFlavor  ProductFlavor  The defaultConfig values
 testVariant  BuildVariant  The Build Variant that will test this variant.
 packageLibrary  Zip  The task that packages the Library AAR archive. Null if not a library.

The TestVariant class adds the following:

 Property Name  Property Type  Description
 buildType  BuildType  The BuildType of the variant.
 productFlavors  List<ProductFlavor>  The ProductFlavors of the variant. Always non Null but could be empty.
 mergedFlavor  ProductFlavor  The merging of android.defaultConfig and variant.productFlavors
 signingConfig  SigningConfig  The SigningConfig object used by this variant
 isSigningReady  boolean  true if the variant has all the information needed to be signed.
 testedVariant  BaseVariant  The BaseVariant that is tested by this TestVariant.
 dex  Dex  The task that dex the code. Can be null if the variant is a library.
 packageApplication  PackageApplication  The task that makes the final APK. Can be null if the variant is a library.
 zipAlign  ZipAlign  The task that zipaligns the apk. Can be null if the variant is a library or if the APK cannot be signed.
 install  DefaultTask  The installation task. Can be null.
 uninstall  DefaultTask  The uninstallation task.
 connectedAndroidTest  DefaultTask  The task that runs the android tests on connected devices.
 providerAndroidTest  DefaultTask  The task that runs the android tests using the extension API.


API for Android specific task types.
  • ProcessManifest
    • File manifestOutputFile
  • AidlCompile
    • File sourceOutputDir
  • RenderscriptCompile
    • File sourceOutputDir
    • File resOutputDir
  • MergeResources
    • File outputDir
  • MergeAssets
    • File outputDir
  • ProcessAndroidResources
    • File manifestFile
    • File resDir
    • File assetsDir
    • File sourceOutputDir
    • File textSymbolOutputDir
    • File packageOutputFile
    • File proguardOutputFile
  • GenerateBuildConfig
    • File sourceOutputDir
  • Dex
    • File outputFolder
  • PackageApplication
    • File resourceFile
    • File dexFile
    • File javaResourceDir
    • File jniDir
    • File outputFile
      • To change the final output file use "outputFile" on the variant object directly.
  • ZipAlign
    • File inputFile
    • File outputFile
      • To change the final output file use "outputFile" on the variant object directly.

The API for each task type is limited due to both how Gradle works and how the Android plugin sets them up.
First, Gradle is meant to have the tasks be only configured for input/output location and possible optional flags. So here, the tasks only define (some of) the inputs/outputs.

Second, the input for most of those tasks is non-trivial, often coming from mixing values from the sourceSets, the Build Types, and the Product Flavors. To keep build files simple to read and understand, the goal is to let developers modify the build by tweak these objects through the DSL, rather than diving deep in the inputs and options of the tasks and changing them.

Also note, that except for the ZipAlign task type, all other types require setting up private data to make them work. This means it’s not possible to manually create new tasks of these types.

This API is subject to change. In general the current API is around giving access to the outputs and inputs (when possible) of the tasks to add extra processing when required). Feedback is appreciated, especially around needs that may not have been foreseen.

For Gradle tasks (DefaultTask, JavaCompile, Copy, Zip), refer to the Gradle documentation.

Setting language level

You can use the compileOptions block to choose the language level used by the compiler. The default is chosen based on the compileSdkVersion value.

android {
    compileOptions {
        sourceCompatibility JavaVersion.VERSION_1_6
        targetCompatibility JavaVersion.VERSION_1_6
    }
}

File naming

Class files Class names are written in UpperCamelCase . For classes that extend an Android component, the name of the class should end ...