FragmentActivity onSaveInstanceState not getting called - android

I have seen a few similar questions about onSaveInstanceState not getting called for Fragments, but in my case Fragments work fine, it's the main FragmentActivity that's having trouble.
The relevant code looks fairly simple:
public class MyFActivity extends FragmentActivity implements ActionBar.TabListener {
String[] allValues; // data to save
#Override
protected void onSaveInstanceState (Bundle outState) {
Log.d("putting it!", allValues.toString());
outState.putStringArray("allValues", allValues);
super.onSaveInstanceState(outState);
}
#Override
protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
if (savedInstanceState != null) {
allValues = savedInstanceState.getStringArray("allValues");
Log.d("getting it!", allValues.toString());
}
}
}
When pausing the activity (using the back button), the onSaveInstanceState is never called, and consequently, savedInstanceState is always null within the onCreate method upon resuming the app. I tried adding a block like this:
#Override
public void onPause() {
super.onPause();
onSaveInstanceState(new Bundle());
}
which was suggested in https://stackoverflow.com/a/14195202/362657 but while onSaveInstanceState then gets called, savedInstanceState remains null within onCreate method. What am I missing?

The issue here is that you are misunderstanding how onSaveInstanceState works. It is designed to save the state of the Activity/Fragment in the case that the OS needs to destroy it for memory reasons or configuration changes. This state is then passed back in onCreate when the Activity/Fragment is returned to / restarted.
In a Fragment, all of their lifecycle callbacks are directly tied to their parent Activity. So onSaveInstanceState gets called on the Fragment when its parent Activity has onSaveInstanceState called.
When pausing the activity (using the back button), the onSaveInstanceState is never called, and consequently, savedInstanceState is always null within the onCreate method upon resuming the app.
When pressing back, the user is destroying the Activity, and therefore its children Fragments, so there is no reason to call onSaveInstanceState, since the instance is being destroyed. When you reopen the Activity, it's a brand new instance, with no saved state, so the Bundle passed in onCreate is null. This is behaving exactly as designed. However, try rotating the device or hitting the home button, then you will see the Activity and its children Fragments have onSaveInstanceState called, and passed back in onCreate when returned to.
The hack you added, directly calling onSaveInstanceState(new Bundle()); inside of onPause, is a very bad practice, as you should never call the lifecycle callbacks directly. Doing so can put your app into illegal states.
If what you really want is for your data to persist beyond an instance of your app, I suggest you look into using SharedPreferences or databases for more advanced data. You can then save your persistent data in onPause() or whenever it changes.

In an update to the accepted answer:
A fragment's onSaveInstanceState may be called if you are using a ViewPager with a FragmentStatePagerAdapter (rather than FragmentPagerAdapter)
FragmentStatePagerAdapter
This version of the pager is more useful when there are a large number of pages, working more like a list view. When pages are not visible to the user, their entire fragment may be destroyed, only keeping the saved state of that fragment. This allows the pager to hold on to much less memory associated with each visited page as compared to FragmentPagerAdapter at the cost of potentially more overhead when switching between pages.
And don't forget:
When using FragmentPagerAdapter the host ViewPager must have a valid ID set.

Not an accurate answer to the question, but may help someone's day.
In my case, I called
#Override
public void onSaveInstanceState(Bundle outState, PersistableBundle outPersistentState) {
super.onSaveInstanceState(outState, outPersistentState);
}
I replaced the above code as below and things worked
#Override
protected void onSaveInstanceState(#NonNull Bundle outState) {
super.onSaveInstanceState(outState);
}

Related

Difference between creating a new fragment and getting an existing one from the manager on rotation

It seems whether I set retainInstance to true or not, when I rotate the device, I get an existing fragment. The difference is that if I set it to true, I get "test = yes!", otherwise I get "test = no!" after rotating the device after clicking the test button to change test. That is, the member variable is kept, if I retain the instance.
But as I have said, even if I do not retain it, I get an existing fragment from the manager, anyway (always get "Reusing existing" on rotation). In that case, if all member variables are lost and the views of the fragment are recreated, what are kept? What is the point of getting an existing instance of the fragment?
In the activity's onCreate,
var frag = supportFragmentManager.findFragmentById(R.id.frame)
if(frag == null)
{
frag = Fragment1.newInstance("", "");
}
else
{
Log.d("sss", "Reusing exsiting");
}
val transaction = supportFragmentManager.beginTransaction()
transaction.replace(R.id.frame, frag)
transaction.commit()
In the fragment,
var test = "no!";
override fun onViewCreated(view: View, savedInstanceState: Bundle?)
{
super.onViewCreated(view, savedInstanceState)
Log.d("sss", "test = " + test);
testButton.setOnClickListener {
test = "yes!";
}
}
I have spent some hours trying to recreate your situation with different scenarios. First of all, I should point out that the life cycle of Fragments are in fact so complicated that during Google I/O 2018, one of the lead developers asked the audiences if onCreate() method of the Activity is invoked first or that of Fragment's. And the answer was that, it depends on the SDK version. But they are focusing more and more on Compatibility Libraries and advice developers to use these to have a universal experience across different devices, as well as enjoying the new APIs that will do the job of dealing with Fragments, so much easier for us.
While looking at the documentation of AppCompatActivity class, I realized that this behavior is their way of dealing with fragments.
Protected methods
void onCreate(Bundle savedInstanceState)
Perform initialization of all fragments.
void onDestroy()
Destroy all fragments.
void onPostCreate(Bundle savedInstanceState)
void onPostResume()
Dispatch onResume() to fragments.
void onSaveInstanceState(Bundle outState)
Save all appropriate fragment state.
void onStart()
Dispatch onStart() to all fragments.
void onStop()
Dispatch onStop() to all fragments.
As you can see they "save all appropriate fragment state" in onSaveInstanceState(); meaning that the states will be restored later on, after the Activity gets destroyed and recreated. So in onDestroy() all fragments get destroyed and when the Activity is created again, they get recreated as well. To make sure, you could override these methods inside both Fragment and Activity and check the result. If you do not check FragmentManager for already attached fragments, onCreate() method of the Fragment will be called twice, once directly by you -adding as a new Fragment- and once by the AppCompatActivity itself.
About the retainInstance = true, the documentation says that it keeps the member variables during configuration changes and will cause a slight difference in the life cycle of the Fragment.
setRetainInstance(true)
Control whether a fragment instance is retained across Activity re-creation (such as from a configuration change). This can only be used with fragments not in the back stack. If set, the fragment lifecycle will be slightly different when an activity is recreated:
onDestroy() will not be called (but onDetach() still will be, because the fragment is being detached from its current activity).
onCreate(Bundle) will not be called since the fragment is not being re-created.
onAttach(Activity) and onActivityCreated(Bundle) will still be called.

Why is Fragment null inside onActivityResult

Here is my setup:
Activity -> FragmentWithPages -> ViewPager{Fragment1, Fragment2}
From Fragment1 I launch a DialogFragment and then from the DialogFragment I launch activityForResult for an implicit camera intent to take picture.
Problem:
Sometimes, when returning from the camera my app crashes inside the onActivityResult of Fragment1. Why does this happen? Now understand the chain of callback of onActivityResult. It would be coming back in the order of Activity.onActivityResult -> FragmentWithPages.onActivityResult -> Fragment1.onActivityResult -> DialogFragment.onActivityResult. So my question is, why is DialogFragment null when I do mDialogFragment.onActivityResult(…)?
I imagine it might have to do with memory: the system kills my app and then restarts it after the Camera app returns. But if that were the case, why is the DialogFragment the broken link in the chain? What can I do to prevent this problem or handle it as if nothing went wrong?
So no I do not mean to simply catch the NPE, that does not really do much as it would void the whole purpose for which I took the picture.
When you return from a startActivityForResult(), there's no guarantee that the activity instance will be the same. The system is free to dispose of the activity that started for result, and reconstruct it to handle the onActivityResult() call.
If this happens, it means of course that any instance fields you had previously assigned will no longer be so. It's up to you to use onSaveInstanceState() and reconstruct your activity instance using the saved instance state Bundle passed in to your onCreate().
You can test this scenario by setting "don't keep activities" in developer settings on your device.
This is a common situation where the system destroys your activity because it needs more resources. You need to implement the onSaveInstanceState method and then restore the instance state in the onCreate or onRestoreInstanceState method.
#Override
public void onSaveInstanceState(Bundle outState) {
super.onSaveInstanceState(outState);
outState.putExtra("yourData", yourParcelableData);
}
#Override
public void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
// keep the fragment and all its data across screen rotation
setRetainInstance(true);
if(savedInstanceState != null){
yourParcelableData = (yourParcelableData)savedInstanceState.getExtra("yourData");
}
}
There is a library that allows you to do it easier.
https://github.com/frankiesardo/icepick

How does `onViewStateRestored` from Fragments work?

I am really confused with the internal state of a Fragment.
I have an Activity holding only one Fragment at once and replaces it, if another Fragment should get shown. From the docs onSaveInstanceState is called ONLY if the Activitys onSaveInstanceState is getting called (which isn't called in my case).
If I stop my Fragment, I'll store its state myself inside a Singleton (yeah, I know I hate Singletons, too, but wasn't my idea to do so).
So I have to recreate the whole ViewHirarchy, create new Views (by using the keyword new), restore its state and return them in onCreateView.
I also have a Checkbox inside this View from which I explicitly do NOT want to store its state.
However the FragmentManager wants to be "intelligent" and calls onViewStateRestored with a Bundle I never created myself, and "restores" the state of the old CheckBox and applies it to my NEW CheckBox. This throws up so many questions:
Can I control the bundle from onViewStateRestored?
How does the FragmentManager take the state of a (probably garbage-collected) CheckBox and applies it to the new one?
Why does it only save the state of the Checkbox (Not of TextViews??)
So to sum it up: How does onViewStateRestored work?
Note I'm using Fragmentv4, so no API > 17 required for onViewStateRestored
Well, sometimes fragments can get a little confusing, but after a while you will get used to them, and learn that they are your friends after all.
If on the onCreate() method of your fragment, you do: setRetainInstance(true); The visible state of your views will be kept, otherwise it won't.
Suppose a fragment called "f" of class F, its lifecycle would go like this:
- When instantiating/attaching/showing it, those are the f's methods that are called, in this order:
F.newInstance();
F();
F.onCreate();
F.onCreateView();
F.onViewStateRestored;
F.onResume();
At this point, your fragment will be visible on the screen.
Assume, that the device is rotated, therefore, the fragment information must be preserved, this is the flow of events triggered by the rotation:
F.onSaveInstanceState(); //save your info, before the fragment is destroyed, HERE YOU CAN CONTROL THE SAVED BUNDLE, CHECK EXAMPLE BELLOW.
F.onDestroyView(); //destroy any extra allocations your have made
//here starts f's restore process
F.onCreateView(); //f's view will be recreated
F.onViewStateRestored(); //load your info and restore the state of f's view
F.onResume(); //this method is called when your fragment is restoring its focus, sometimes you will need to insert some code here.
//store the information using the correct types, according to your variables.
#Override
public void onSaveInstanceState(Bundle outState) {
super.onSaveInstanceState(outState);
outState.putSerializable("foo", this.foo);
outState.putBoolean("bar", true);
}
#Override
public void onViewStateRestored(Bundle inState) {
super.onViewStateRestored(inState);
if(inState!=null) {
if (inState.getBoolean("bar", false)) {
this.foo = (ArrayList<HashMap<String, Double>>) inState.getSerializable("foo");
}
}
}

Android: using savedInstanceState with fragments

Have a question regarding the savedInstanceState bundle that you get on the fragment's callbacks when the fragment is first created and then reattached to the activity.
So I use the setRetainInstance(true), as a result the fragment shouldn't be destroyed but just unattached from the activity when the activity is destroyed and then reattached back when the activity is recreated on a configuration change for example.
So, because using this setRetainInstance(true) this will cause the savedInstanceState bundle to ALWAYS be null in the fragment's callbacks such as: onActivityCreated(), onCreate() etc.
So far so good, now I'm getting some crashes from some users (a really really small number) which is caused by this savedInstanceState not being null on the fragment.
So, the onActivityCreated() callback is doing something like this:
#Override
public void onActivityCreated(Bundle savedInstanceState) {
super.onActivityCreated(savedInstanceState);
adapter = new CustomAdapter(getActivity(), getListView(), data, savedInstanceState);
setListAdapter(adapter);
setupEmptyListView();
getListView().setOnScrollListener(this);
}
The CustomAdapter is calling at some point from its constructor a loadInstanceState() method which does its job, code below:
#Override
public void loadInstanceState(Bundle savedInstanceState) {
if (savedInstanceState != null) {
//do some stuff which is causing a crash
}
}
Now my question is how is it possible to get into that if statement when the savedInstanceState that I'm passing should always null.
And now the QUESTION:
Is Android guaranteeing THAT WHEN USING setRetainInstance(true) on a fragment the savedInstanceState bundle which is passed to the callbacks will ALWAYS be null?
Thanks for answers guys!
I didn't find any mention that savedInstanceState bundle must be null in this case.
The official documentation says:
public void setRetainInstance (boolean retain)
Control whether a fragment instance is retained
across Activity re-creation (such as from a configuration change).
This can only be used with fragments not in the back stack. If set,
the fragment lifecycle will be slightly different when an activity is
recreated:
onDestroy() will not be called (but onDetach() still will be, because
the fragment is being detached from its current activity).
onCreate(Bundle) will not be called since the fragment is not being
re-created.
onAttach(Activity) and onActivityCreated(Bundle) will
still be called.
IMHO that's main difference - changed lifecycle.

Duty of savedInstanceState in Android [duplicate]

I'm trying to save and restore the state of an Activity using the methods onSaveInstanceState() and onRestoreInstanceState().
The problem is that it never enters the onRestoreInstanceState() method. Can anyone explain to me why this is?
Usually you restore your state in onCreate(). It is possible to restore it in onRestoreInstanceState() as well, but not very common. (onRestoreInstanceState() is called after onStart(), whereas onCreate() is called before onStart().
Use the put methods to store values in onSaveInstanceState():
protected void onSaveInstanceState(Bundle icicle) {
super.onSaveInstanceState(icicle);
icicle.putLong("param", value);
}
And restore the values in onCreate():
public void onCreate(Bundle icicle) {
if (icicle != null){
value = icicle.getLong("param");
}
}
onRestoreInstanceState() is called only when recreating activity after it was killed by the OS. Such situation happen when:
orientation of the device changes (your activity is destroyed and recreated).
there is another activity in front of yours and at some point the OS kills your activity in order to free memory (for example). Next time when you start your activity onRestoreInstanceState() will be called.
In contrast: if you are in your activity and you hit Back button on the device, your activity is finish()ed (i.e. think of it as exiting desktop application) and next time you start your app it is started "fresh", i.e. without saved state because you intentionally exited it when you hit Back.
Other source of confusion is that when an app loses focus to another app onSaveInstanceState() is called but when you navigate back to your app onRestoreInstanceState() may not be called. This is the case described in the original question, i.e. if your activity was NOT killed during the period when other activity was in front onRestoreInstanceState() will NOT be called because your activity is pretty much "alive".
All in all, as stated in the documentation for onRestoreInstanceState():
Most implementations will simply use onCreate(Bundle) to restore their
state, but it is sometimes convenient to do it here after all of the
initialization has been done or to allow subclasses to decide whether
to use your default implementation. The default implementation of this
method performs a restore of any view state that had previously been
frozen by onSaveInstanceState(Bundle).
As I read it: There is no reason to override onRestoreInstanceState() unless you are subclassing Activity and it is expected that someone will subclass your subclass.
The state you save at onSaveInstanceState() is later available at onCreate() method invocation. So use onCreate (and its Bundle parameter) to restore state of your activity.
From the documentation Restore activity UI state using saved instance state it is stated as:
Instead of restoring the state during onCreate() you may choose to
implement onRestoreInstanceState(), which the system calls after the
onStart() method. The system calls onRestoreInstanceState() only if
there is a saved state to restore, so you do not need to check whether
the Bundle is null:
IMO, this is more clear way than checking this at onCreate, and better fits with single responsiblity principle.
As a workaround, you could store a bundle with the data you want to maintain in the Intent you use to start activity A.
Intent intent = new Intent(this, ActivityA.class);
intent.putExtra("bundle", theBundledData);
startActivity(intent);
Activity A would have to pass this back to Activity B. You would retrieve the intent in Activity B's onCreate method.
Intent intent = getIntent();
Bundle intentBundle;
if (intent != null)
intentBundle = intent.getBundleExtra("bundle");
// Do something with the data.
Another idea is to create a repository class to store activity state and have each of your activities reference that class (possible using a singleton structure.) Though, doing so is probably more trouble than it's worth.
The main thing is that if you don't store in onSaveInstanceState() then onRestoreInstanceState() will not be called. This is the main difference between restoreInstanceState() and onCreate(). Make sure you really store something. Most likely this is your problem.
I found that onSaveInstanceState is always called when another Activity comes to the foreground. And so is onStop.
However, onRestoreInstanceState was called only when onCreate and onStart were also called. And, onCreate and onStart were NOT always called.
So it seems like Android doesn't always delete the state information even if the Activity moves to the background. However, it calls the lifecycle methods to save state just to be safe. Thus, if the state is not deleted, then Android doesn't call the lifecycle methods to restore state as they are not needed.
Figure 2 describes this.
I think this thread was quite old. I just mention another case, that onSaveInstanceState() will also be called, is when you call Activity.moveTaskToBack(boolean nonRootActivity).
If you are handling activity's orientation changes with android:configChanges="orientation|screenSize" and onConfigurationChanged(Configuration newConfig), onRestoreInstanceState() will not be called.
It is not necessary that onRestoreInstanceState will always be called after onSaveInstanceState.
Note that :
onRestoreInstanceState will always be called, when activity is rotated (when orientation is not handled) or open your activity and then open other apps so that your activity instance is cleared from memory by OS.
In my case, onRestoreInstanceState was called when the activity was reconstructed after changing the device orientation. onCreate(Bundle) was called first, but the bundle didn't have the key/values I set with onSaveInstanceState(Bundle).
Right after, onRestoreInstanceState(Bundle) was called with a bundle that had the correct key/values.
I just ran into this and was noticing that the documentation had my answer:
"This function will never be called with a null state."
https://developer.android.com/reference/android/view/View.html#onRestoreInstanceState(android.os.Parcelable)
In my case, I was wondering why the onRestoreInstanceState wasn't being called on initial instantiation. This also means that if you don't store anything, it'll not be called when you go to reconstruct your view.
I can do like that (sorry it's c# not java but it's not a problem...) :
private int iValue = 1234567890;
function void MyTest()
{
Intent oIntent = new Intent (this, typeof(Camera2Activity));
Bundle oBundle = new Bundle();
oBundle.PutInt("MYVALUE", iValue); //=> 1234567890
oIntent.PutExtras (oBundle);
iRequestCode = 1111;
StartActivityForResult (oIntent, 1111);
}
AND IN YOUR ACTIVITY FOR RESULT
private int iValue = 0;
protected override void OnCreate(Bundle bundle)
{
Bundle oBundle = Intent.Extras;
if (oBundle != null)
{
iValue = oBundle.GetInt("MYVALUE", 0);
//=>1234567890
}
}
private void FinishActivity(bool bResult)
{
Intent oIntent = new Intent();
Bundle oBundle = new Bundle();
oBundle.PutInt("MYVALUE", iValue);//=>1234567890
oIntent.PutExtras(oBundle);
if (bResult)
{
SetResult (Result.Ok, oIntent);
}
else
SetResult(Result.Canceled, oIntent);
GC.Collect();
Finish();
}
FINALLY
protected override void OnActivityResult(int iRequestCode, Android.App.Result oResultCode, Intent oIntent)
{
base.OnActivityResult (iRequestCode, oResultCode, oIntent);
iValue = oIntent.Extras.GetInt("MYVALUE", -1); //=> 1234567890
}

Categories

Resources