For example, let's say my Rest adapter created with Retrofit lives inside Application class.
I would love to get it inside the Activity, so I write the following code:
public class MainActivity extends Activity {
#Inject MyRestAdapter mRestAdapter;
#Override
public void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
((GlobalApplication) getApplication()).getComponent().inject(this);
}
}
Granted, it will make the job done. But...
How is this different from calling getApplication(), and then explicitly yank the MyRestAdapter to MainActivity? Yes, Dagger 2 will simplify the setup by automatically getting everything to the Activity, but you still need to explicitly tell from where you need these dependencies, and that, if I understand correctly, defeats the whole purpose of DI. Am I right to say that Dagger 2 is "semi-automated service locator", or it's just the tutorials that misled me, and there is correct way to inject dependencies with Dagger 2 into the View or Activity from Application?
I've been experimenting with Dagger and it definitely seems to blur the lines between service locator and dependency injection. This is at least true when used with Android activities. With the current version of Dagger, it is possible to write AndroidInjection.inject(this) in an activity's onCreate method. That's basically like saying "find all the services I need and inject them into me." So Dagger is a combination of a some central/global service locator that knows where to get the services and an injector that knows where (i.e. which instance variables) to put those services in the activity. It seems that Android activities force reliance on some kind of singleton/global object.
There is no 1 "purpose of DI" but definitely one of them is to separate the configuration from what actually requests the configured objects. The idea is that your higher-level objects, such as Activitys in Android, can request all of the objects it needs without worrying about where they come from, how they are constructed, and any semantics about their relationships. Similar to how an Activity doesn't deal with drawing text to the screen (and instead delegates that to a TextView[0]), DI helps keep your objects from knowing too much that is not relevant to the actual logic they need to perform.
Inherent in the "semi-automated service locator" you describe is static analysis and error handling. As applications become larger, it becomes even harder to get manual-DI correct. Dagger helps make your code less error-prone (and less tedious to maintain).
Consider the case where you have an internal version of your app for employees, where you log lots of information about how they use the app to make sure you can identify any issues. In your actual product, however, you don't want to track personally identifiable information like that when it's not necessary. Now, your MainActivity needs an AnalyticsLogger - which one should it get? The more cases you have, the easier it is for Dagger to help piece things together than for you to do it yourself.
[0] which delegates to a Paint object
This is more of a question where I'm trying to see if my thinking is correct before I start coding something that winds up not working. Let me see if I can explain what I'm trying to do first. My main issue is wanting to know if I'm on the right track or not.
I'm creating a game using the page viewer and fragments so I can have a tabbed layout. That's not a problem. Where I'm getting confused and turned around is that I have multiple activities, each handling a different part of the game so that I avoid having too much on the screen at any one time. I had wanted to have all of the game methods in a single class and have each activity pull from it. Can I do so if the class is private? I assume if I had a GAME class for example that I would have to instantiate a particular version of it somewhere but if I did, how can I be sure that everything is pulling from that same instance?
Another approach I had thought of was basically just making one game class that did everything and then just spit out the variable values to the various activities as needed which would then update the UI. Would that be a better approach? I think if I do it this way I can avoid having to pass data directly from one activity to the next. I can just have them all pull whatever values they need directly from the class.
Does any of this make any sense or am I way out in left field somewhere? Any direction to go in to start would be helpful as I've been going around for days trying to chart some kind of course.
It sounds like you want to put all your game logic into a Game class and have all of your Activities access that single Game instance. The answer is yes, that's probably the right approach and a simpler approach then trying to get your Activities to pass data around via Intents.
You can use a singleton pattern to get access to the same instance everywhere. (Something like this)
public class Game {
private static final Game INSTANCE = new Game();
private Game() {}
public static Game getInstance() {
return INSTANCE;
}
}
Or if you're using a dependency injection framework like Dagger, you can just mark the Game provider as providing a singleton. (If you don't know what this is, just use the singleton pattern above for now)
I'm working on an application that has multiple Activities, and I'm tired of running back and forth between the Manifest and XML layout and stuff. Is there a way to have
Intent intent = new Intent(MainActivity.this, MainActivity.Settings.class);
Or something? Because I've tried it, it doesn't throw me an error, but it just force closes the application. I'm able to bundle all my classes from different .java into one, for ex.
public class MainActivity extends Activity
{
...
#Override
protected void onCreate(Bundle MainActivityState)
{
...
}
public class Settings extends ...
{
...
}
public class Register extends ...
{
...
}
public class Login extends ...
{
...
}
public class BeautifulLady extends personality ...
}
Simple. Just don't even try.
An activity loosely represents a single screen - something the user interacts with. Android is built around this concept and trying to circumvent it will lead to tears.
Stick with it. Having your classes in separate files, and having layout XML separate for each activity, will become your friend and will actually speed things up once you are familiar.
Start with the Activity life cycle document and read it several times until the penny drops. Then expand out from there.
http://developer.android.com/reference/android/app/Activity.html
Object oriented programming, with classes that take care of themselves, is a joy and regardless of which platform you choose to develop on is the way to go for the foreseeable future (old hands, no debates on OOP vs functional please ;)).
If you are going to do mobile development, then the separation of activities, classes and UI is the same concept, just done differently.
See also MVC programming and its' cousins.
http://en.wikipedia.org/wiki/Model%E2%80%93view%E2%80%93controller
Good luck.
Perhaps you can define your Activity as 'Single Top', then launch your activity from herself like MainActivity.this.startActivity(new Intent(MainActivity.this, MainActivity.class). It'll then go into onNewIntent() and you will redisplay what you want to redisplay. This way you will have only one screen.
Android newbee here, I have some code that I want to run when my android app first starts up. It checks the version of the local database and downloads a new version if the current version is out of date. I have been sticking it in the oncreate of my first activity, pretty sure there has to be a better place to put this. Any recommendations of somewhere I can put it where it will get called once on startup?
You can write a custom Application class (extend from android.app.Application). Override onCreate to specify what happens when the application is started:
public class MyApplication extends Application {
#Override
public void onCreate() {
super.onCreate();
// Do something here.
}
}
You'll then need to register your custom class in the manifest file:
<application ... android:name="fully.qualified.MyApplication">
Edit:
In response to David Cesarino, I disagree with the purpose of the Application class. If you rely on the Activity's onCreate, then what's to stop it from becoming the same huge class of miscellaneous purposes... if you need something to happen when the application starts, you have to write that code somewhere; and the Activity would probably become more cluttered because you have to perform Activity specific logic in it as well. If you're worried about clutter, then separate the logic into other classes and call them from the Application. Using the SharedPreferences to determine whether or not the logic should execute seems like more of a work-around to a problem that's already been solved.
Dianne Hackborn seems to be referring to data, not logic, in which I totally agree. Static variables are much better than Application level variables... better scoping and type safety make maintainability/readability much easier.
First, look at the Activity lifecycle.
Answering your question, you could put code in any of those "start-up" methods, depending on what you want to do and, mostly important, when you want to trigger that. For what you asked, onCreate is the reasonable place.
I have been sticking it in the oncreate of my first activity, pretty sure there has to be a better place to put this.
And why is that? Any code has an entry point, right? In Android Activities it just happens to be onCreate (again, see above link for the full details). Besides event handling, which are responses to events happening outside the main sequence of calls, you put stuff in onCreate.
If you're concerned about the method becoming huge, then that's another problem. Abstract your code better, I say. For checking preliminary stuff, people generally provide a "Loading" activity, before starting the main activity of the app.
edited:
This is a follow up to what drumboog proposed, since my comment started to grow in complexity to be "just a comment".
Personally, I'd avoid extending the Application class for the sole reason of executing code early on, more so a code that is not that sensible in priority (versioning databases). The Application class is mostly used as an easy way to persist state between Activity'ies, not as a way to "do everything". In short, I feel the Application class is commonly abused.
For what you want, you could perfectly achieve that calling code in Activity onCreate. That reduces complexity, because I've seen people stuffing Application until it becomes a huge class of miscellaneous code purposes. And that's a no-no for maintenance, with logic problems of its own.
Besides, if you truly want another solution, completely disassociated with the UI, you should think about implementing a Service instead (but I don't think it's necessary for just that).
Both of those concerns were previously addressed by Dianne Hackborn (or what I got from her message).
I am creating an application which requires login. I created the main and the login activity.
In the main activity onCreate method I added the following condition:
public void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
setContentView(R.layout.main);
...
loadSettings();
if(strSessionString == null)
{
login();
}
...
}
The onActivityResult method which is executed when the login form terminates looks like this:
#Override
public void onActivityResult(int requestCode,
int resultCode,
Intent data)
{
super.onActivityResult(requestCode, resultCode, data);
switch(requestCode)
{
case(SHOW_SUBACTICITY_LOGIN):
{
if(resultCode == Activity.RESULT_OK)
{
strSessionString = data.getStringExtra(Login.SESSIONSTRING);
connectionAvailable = true;
strUsername = data.getStringExtra(Login.USERNAME);
}
}
}
The problem is the login form sometimes appears twice (the login() method is called twice) and also when the phone keyboard slides the login form appears again and I guess the problem is the variable strSessionString.
Does anyone know how to set the variable global in order to avoid login form appearing after the user already successfully authenticates?
I wrote this answer back in '09 when Android was relatively new, and there were many not well established areas in Android development. I have added a long addendum at the bottom of this post, addressing some criticism, and detailing a philosophical disagreement I have with the use of Singletons rather than subclassing Application. Read it at your own risk.
ORIGINAL ANSWER:
The more general problem you are encountering is how to save state across several Activities and all parts of your application. A static variable (for instance, a singleton) is a common Java way of achieving this. I have found however, that a more elegant way in Android is to associate your state with the Application context.
As you know, each Activity is also a Context, which is information about its execution environment in the broadest sense. Your application also has a context, and Android guarantees that it will exist as a single instance across your application.
The way to do this is to create your own subclass of android.app.Application, and then specify that class in the application tag in your manifest. Now Android will automatically create an instance of that class and make it available for your entire application. You can access it from any context using the Context.getApplicationContext() method (Activity also provides a method getApplication() which has the exact same effect). Following is an extremely simplified example, with caveats to follow:
class MyApp extends Application {
private String myState;
public String getState(){
return myState;
}
public void setState(String s){
myState = s;
}
}
class Blah extends Activity {
#Override
public void onCreate(Bundle b){
...
MyApp appState = ((MyApp)getApplicationContext());
String state = appState.getState();
...
}
}
This has essentially the same effect as using a static variable or singleton, but integrates quite well into the existing Android framework. Note that this will not work across processes (should your app be one of the rare ones that has multiple processes).
Something to note from the example above; suppose we had instead done something like:
class MyApp extends Application {
private String myState = /* complicated and slow initialization */;
public String getState(){
return myState;
}
}
Now this slow initialization (such as hitting disk, hitting network, anything blocking, etc) will be performed every time Application is instantiated! You may think, well, this is only once for the process and I'll have to pay the cost anyways, right? For instance, as Dianne Hackborn mentions below, it is entirely possible for your process to be instantiated -just- to handle a background broadcast event. If your broadcast processing has no need for this state you have potentially just done a whole series of complicated and slow operations for nothing. Lazy instantiation is the name of the game here. The following is a slightly more complicated way of using Application which makes more sense for anything but the simplest of uses:
class MyApp extends Application {
private MyStateManager myStateManager = new MyStateManager();
public MyStateManager getStateManager(){
return myStateManager ;
}
}
class MyStateManager {
MyStateManager() {
/* this should be fast */
}
String getState() {
/* if necessary, perform blocking calls here */
/* make sure to deal with any multithreading/synchronicity issues */
...
return state;
}
}
class Blah extends Activity {
#Override
public void onCreate(Bundle b){
...
MyStateManager stateManager = ((MyApp)getApplicationContext()).getStateManager();
String state = stateManager.getState();
...
}
}
While I prefer Application subclassing to using singletons here as the more elegant solution, I would rather developers use singletons if really necessary over not thinking at all through the performance and multithreading implications of associating state with the Application subclass.
NOTE 1: Also as anticafe commented, in order to correctly tie your Application override to your application a tag is necessary in the manifest file. Again, see the Android docs for more info. An example:
<application
android:name="my.application.MyApp"
android:icon="..."
android:label="...">
</application>
NOTE 2: user608578 asks below how this works with managing native object lifecycles. I am not up to speed on using native code with Android in the slightest, and I am not qualified to answer how that would interact with my solution. If someone does have an answer to this, I am willing to credit them and put the information in this post for maximum visibility.
ADDENDUM:
As some people have noted, this is not a solution for persistent state, something I perhaps should have emphasized more in the original answer. I.e. this is not meant to be a solution for saving user or other information that is meant to be persisted across application lifetimes. Thus, I consider most criticism below related to Applications being killed at any time, etc..., moot, as anything that ever needed to be persisted to disk should not be stored through an Application subclass. It is meant to be a solution for storing temporary, easily re-creatable application state (whether a user is logged in for example) and components which are single instance (application network manager for example) (NOT singleton!) in nature.
Dayerman has been kind enough to point out an interesting conversation with Reto Meier and Dianne Hackborn in which use of Application subclasses is discouraged in favor of Singleton patterns. Somatik also pointed out something of this nature earlier, although I didn't see it at the time. Because of Reto and Dianne's roles in maintaining the Android platform, I cannot in good faith recommend ignoring their advice. What they say, goes. I do wish to disagree with the opinions, expressed with regards to preferring Singleton over Application subclasses. In my disagreement I will be making use of concepts best explained in this StackExchange explanation of the Singleton design pattern, so that I do not have to define terms in this answer. I highly encourage skimming the link before continuing. Point by point:
Dianne states, "There is no reason to subclass from Application. It is no different than making a singleton..." This first claim is incorrect. There are two main reasons for this. 1) The Application class provides a better lifetime guarantee for an application developer; it is guaranteed to have the lifetime of the application. A singleton is not EXPLICITLY tied to the lifetime of the application (although it is effectively). This may be a non-issue for your average application developer, but I would argue this is exactly the type of contract the Android API should be offering, and it provides much more flexibility to the Android system as well, by minimizing the lifetime of associated data. 2) The Application class provides the application developer with a single instance holder for state, which is very different from a Singleton holder of state. For a list of the differences, see the Singleton explanation link above.
Dianne continues, "...just likely to be something you regret in the future as you find your Application object becoming this big tangled mess of what should be independent application logic." This is certainly not incorrect, but this is not a reason for choosing Singleton over Application subclass. None of Diane's arguments provide a reason that using a Singleton is better than an Application subclass, all she attempts to establish is that using a Singleton is no worse than an Application subclass, which I believe is false.
She continues, "And this leads more naturally to how you should be managing these things -- initializing them on demand." This ignores the fact that there is no reason you cannot initialize on demand using an Application subclass as well. Again there is no difference.
Dianne ends with "The framework itself has tons and tons of singletons for all the little shared data it maintains for the app, such as caches of loaded resources, pools of objects, etc. It works great." I am not arguing that using Singletons cannot work fine or are not a legitimate alternative. I am arguing that Singletons do not provide as strong a contract with the Android system as using an Application subclass, and further that using Singletons generally points to inflexible design, which is not easily modified, and leads to many problems down the road. IMHO, the strong contract the Android API offers to developer applications is one of the most appealing and pleasing aspects of programming with Android, and helped lead to early developer adoption which drove the Android platform to the success it has today. Suggesting using Singletons is implicitly moving away from a strong API contract, and in my opinion, weakens the Android framework.
Dianne has commented below as well, mentioning an additional downside to using Application subclasses, they may encourage or make it easier to write less performance code. This is very true, and I have edited this answer to emphasize the importance of considering perf here, and taking the correct approach if you're using Application subclassing. As Dianne states, it is important to remember that your Application class will be instantiated every time your process is loaded (could be multiple times at once if your application runs in multiple processes!) even if the process is only being loaded for a background broadcast event. It is therefore important to use the Application class more as a repository for pointers to shared components of your application rather than as a place to do any processing!
I leave you with the following list of downsides to Singletons, as stolen from the earlier StackExchange link:
Inability to use abstract or interface classes;
Inability to subclass;
High coupling across the application (difficult to modify);
Difficult to test (can't fake/mock in unit tests);
Difficult to parallelize in the case of mutable state (requires extensive locking);
and add my own:
Unclear and unmanageable lifetime contract unsuited for Android (or most other) development;
Create this subclass
public class MyApp extends Application {
String foo;
}
In the AndroidManifest.xml add android:name
Example
<application android:name=".MyApp"
android:icon="#drawable/icon"
android:label="#string/app_name">
The suggested by Soonil way of keeping a state for the application is good, however it has one weak point - there are cases when OS kills the entire application process. Here is the documentation on this - Processes and lifecycles.
Consider a case - your app goes into the background because somebody is calling you (Phone app is in the foreground now). In this case && under some other conditions (check the above link for what they could be) the OS may kill your application process, including the Application subclass instance. As a result the state is lost. When you later return to the application, then the OS will restore its activity stack and Application subclass instance, however the myState field will be null.
AFAIK, the only way to guarantee state safety is to use any sort of persisting the state, e.g. using a private for the application file or SharedPrefernces (it eventually uses a private for the application file in the internal filesystem).
Just a note ..
add:
android:name=".Globals"
or whatever you named your subclass to the existing <application> tag. I kept trying to add another <application> tag to the manifest and would get an exception.
What about ensuring the collection of native memory with such global structures?
Activities have an onPause/onDestroy() method that's called upon destruction, but the Application class has no equivalents. What mechanism are recommended to ensure that global structures (especially those containing references to native memory) are garbage collected appropriately when the application is either killed or the task stack is put in the background?
I couldn't find how to specify the application tag either, but after a lot of Googling, it became obvious from the manifest file docs: use android:name, in addition to the default icon and label in the application stanza.
android:name
The fully qualified name of an Application subclass implemented for the application. When the application process is started, this class is instantiated before any of the application's components.
The subclass is optional; most applications won't need one. In the absence of a subclass, Android uses an instance of the base Application class.
Just you need to define an application name like below which will work:
<application
android:name="ApplicationName" android:icon="#drawable/icon">
</application>
Like there was discussed above OS could kill the APPLICATION without any notification (there is no onDestroy event) so there is no way to save these global variables.
SharedPreferences could be a solution EXCEPT you have COMPLEX STRUCTURED variables (in my case I had integer array to store the IDs that the user has already handled). The problem with the SharedPreferences is that it is hard to store and retrieve these structures each time the values needed.
In my case I had a background SERVICE so I could move this variables to there and because the service has onDestroy event, I could save those values easily.
If some variables are stored in sqlite and you must use them in most activities in your app.
then Application maybe the best way to achieve it.
Query the variables from database when application started and store them in a field.
Then you can use these variables in your activities.
So find the right way, and there is no best way.
You can have a static field to store this kind of state. Or put it to the resource Bundle and restore from there on onCreate(Bundle savedInstanceState). Just make sure you entirely understand Android app managed lifecycle (e.g. why login() gets called on keyboard orientation change).
DO N'T Use another <application> tag in manifest file.Just do one change in existing <application> tag , add this line android:name=".ApplicationName" where, ApplicationName will be name of your subclass(use to store global) that, you is about to create.
so, finally your ONE AND ONLY <application> tag in manifest file should look like this :-
<application
android:allowBackup="true"
android:icon="#mipmap/ic_launcher"
android:label="#string/app_name"
android:theme="#style/Theme.AppCompat.NoActionBar"
android:name=".ApplicationName"
>
you can use Intents , Sqlite , or Shared Preferences . When it comes to the media storage, like documents , photos , and videos, you may create the new files instead.
You can do this using two approaches:
Using Application class
Using Shared Preferences
Using Application class
Example:
class SessionManager extends Application{
String sessionKey;
setSessionKey(String key){
this.sessionKey=key;
}
String getSessisonKey(){
return this.sessionKey;
}
}
You can use above class to implement login in your MainActivity as below. Code will look something like this:
#override
public void onCreate (Bundle savedInstanceState){
// you will this key when first time login is successful.
SessionManager session= (SessionManager)getApplicationContext();
String key=getSessisonKey.getKey();
//Use this key to identify whether session is alive or not.
}
This method will work for temporary storage. You really do not any idea when operating system is gonna kill the application, because of low memory.
When your application is in background and user is navigating through other application which demands more memory to run, then your application will be killed since operating system given more priority to foreground processes than background.
Hence your application object will be null before user logs out. Hence for this I recommend to use second method Specified above.
Using shared preferences.
String MYPREF="com.your.application.session"
SharedPreferences pref= context.getSharedPreferences(MyPREF,MODE_PRIVATE);
//Insert key as below:
Editot editor= pref.edit();
editor.putString("key","value");
editor.commit();
//Get key as below.
SharedPreferences sharedPref = getActivity().getPreferences(Context.MODE_PRIVATE);
String key= getResources().getString("key");
On activity result is called before on resume. So move you login check to on resume and your second login can be blocked once the secomd activity has returned a positive result. On resume is called every time so there is not worries of it not being called the first time.
The approach of subclassing has also been used by the BARACUS framework. From my point of view subclassing Application was intended to work with the lifecycles of Android; this is what any Application Container does. Instead of having globals then, I register beans to this context an let them beeing injected into any class manageable by the context. Every injected bean instance actually is a singleton.
See this example for details
Why do manual work if you can have so much more?
class GlobaleVariableDemo extends Application {
private String myGlobalState;
public String getGlobalState(){
return myGlobalState;
}
public void setGlobalState(String s){
myGlobalState = s;
}
}
class Demo extends Activity {
#Override
public void onCreate(Bundle b){
...
GlobaleVariableDemo appState = ((GlobaleVariableDemo)getApplicationContext());
String state = appState.getGlobalState();
...
}
}
You could create a class that extends Application class and then declare your variable as a field of that class and providing getter method for it.
public class MyApplication extends Application {
private String str = "My String";
synchronized public String getMyString {
return str;
}
}
And then to access that variable in your Activity, use this:
MyApplication application = (MyApplication) getApplication();
String myVar = application.getMyString();