I have two activities A and B. The A has a ListFragment which uses LoaderManager, whereas B activity shows a details about the item selected in the A's ListFragment. I've just noticed that when I use a back button to get from the B back to the A, the position in the ListFragment preserve, but when I use the up button (left-point caret) in the action bar, the A activity is recreated and thus position in list view is lost.
I would like fix this issue, but I am not sure about the best way how to do it right.
I come up with this solutions:
a) Use onBackPressed()
Replace the default implementation for the android.R.id.home (the up action bar button) in the B activity, and instead of the NavUtils.navigateUpFromSameTask(this) function call the onBackPressed() activity method. I've tested it and it works.
b) Keep use NavUtils.navigateUpFromSameTask(this)
But implement the onSaveInstanceState and restore listView position during onCreate method of the ListFragment used by the A activity. (I've not tested this approach yet)
Which of this solutions is better? Or is there any other (much more better) solution?
Solution a) is pretty simple and straight forward, but b) is probably better because the default implementation of the up caret is used.
Any ideas are welcome. Thanks.
Solution c is the correct option. First, though, an explanation of the problem with solution a.
There is absolutely no point in having two back buttons in your Activity. Furthermore, option a actually breaks the up button. The point of the up button is to provide a way for users to stay within your app when they have landed in your app from an outside source. For example, if you land on activity B from an outside activity C and if you are using your option a, then pressing "up" in activity B will result in activity C being shown. The behavior you would want would be for activity A to be shown.
As you can see, solution b is on the right track. You definitely want to go up to A and not back to C. However, simply storing the state in onSaveInstanceState will not cause the state to be retained. This is because onSaveInstanceState only gets called if your application may be killed by the system. It is not guaranteed to be called if your application was destroyed manually, and it certainly won't be called when a new instance of your Activity is created. If the Intent starts a new activity, then it will not have its state restored from the other activity.
There solution, then, is that you must write anything persistent to a shared preference file (or a custom persistent alternative). When doing this you can guarantee that all instances of an Activity share the same state across multiple tasks so long as their onResume (or wherever you restore state) is called. OR:
If you know exactly how you want your navigation to work, you can avoid writing everything to persistent state by using a combination of Intent flags and Activity task affinities. If you want to use the same activity as up even if you navigate into the application from an outside source, then you can leave your Activity A's affinity as default (linked to the application) and use something like Intent.FLAG_ACTIVITY_CLEAR_TOP.
Personally, I'd try the Intent flag approach first and failing that fall back to writing the state persistently. You just don't really want scroll location sitting on persistent storage if you can avoid it..
check out this presentation: https://speakerdeck.com/jgilfelt/this-way-up-implementing-effective-navigation-on-android. It answers to all of your problems.
Related
I have an activity A. Activity B is partially transparent (so I can see activity A below).
Is there an option in Android to allow a user to interact with Activity A while B is still in front?
I want to leave activity B in front the whole time, but allow the user to interact with screen A.
overriding the implementation of onTouch or onClick in B so I can pass the event to activity A?
This is not possible and will not be, because it's a security concern. This on web is called clickjacking. The reason that this is not possible is because you could theoretically manipulate the user to do something unintended like deleting a file or similar, just by guiding them on where to click on a fake screen.
in my experience there is nothing like this. So, you must close activity A and B can be clickable or something.
I know that in Android, if stuff is idle for a while, the operating system will devour things to free up memory.
So if I have a first Activity, and I invoke a second Activity by using an Intent, and then invoke a third Activity using yet another Intent, I can use the back button to go back to the previous Activities.
But let's say I stay on the third Activity, and let the phone idle for a while until the OS decides to devour my app for memory. If I open the app again, will I have lost the stack I have formed from my Intents? Will I still be in the third Activity with the ability to press Back and go to Activity 2, then Activity 1?
The OS will handle how long a particular Activity stays "active" in memory. However, the "stack" shouldn't change, regardless of whether an activity is "active" or not. This is where the Bundle comes in handy and the methods: "onSaveInstanceState()" and "onRestoreInstanceState()".
Implementing these methods properly is the difference of the activity reappearing on the screen in an empty state vs. with its previous state maintained.
Some documentation on Recreating an Activity
The backstack stays intact for the task in hand. It will always stay in task and whenever the user presses the back button, it will go through the back stack like popping the last item. However, not all the activities in the stack are in the foreground. Usually, only the last item put into the back stack (the top of the stack) is in the foreground and if there are multiple apps/tasks open this may not even be true. Here is a great diagram to show this.
Now, lets say a user opens a task with a couple of activities in its back stack. The top activity is in the foreground and is running normally, but to preserve memory the other activities were destroyed. So now when the user presses the back button, the task knows what activity was in the back stack and knows it is now destroyed. So, it will recreate it following the Activities lifecycle and any data that was in it will be lost. One way to preserve it (mentioned by original answer) is using the onSaveInstanceState() and onRestoreInstanceState(), which save things in a bundle that Android preserves for the user so data can be saved. All of this information can be found in the docs. To answer your question more clearly, yes you will be in the activity you think you would be in, but you can treat it as a new instance of that activity and to recover the data from before to display it in the same way, you should use bundle and implement the methods aforementioned.
Situation:
Starting from my host activityA/fragment A, I click on a button to start host activityB/fragment B. Fragment B is a fragment which enables filtering options for Fragment A. After selecting the options, the user can press the navigate up button to return to fragment A.
Problem:
I would like the state of the options selected to be retained when navigating to Fragment B more than once. Since this is a filtering option, it would be preferable to not save this in storage beyond the duration of the application (It's just filtering. Not necessary to keep the information stored for a long time. Just in between navigation).
Things I've tried:
OnSavedInstanceState - realized that onNavigateUpTo() / finish() don't trigger onSavedInstanceState
android:launchmode="singleTop" - Because the activity is finished and destroyed from the backstack, no instance of the activity is available to receive the new intent.
setRetainInstance - Activity is destroyed so attached fragment also destroyed.
A possible solution I've found is to use setResult and return the values to Fragment A. Then put these values as EXTRAs into a new intent when starting Fragment B again. This solution seems clunky. Is there a cleaner solution that just allows Fragment B to "remember" its state instead of passing values back and forth?
As you have stated the fragment is destroyed when user goes back so there is no way to get the "state" back. Also it does not make sense to store state as logically the fragment/activity is not needed from platform point of view.
So the way to 'remember' the state of fragment is to store the filter data in some other variable(s) and give it back to it when it is re-launched. What you are trying to do it perfectly fine i.e. return the filter data as result and send it back to fragment at re-launch.
To make it simpler you may write a class containing all the filter options and make it Parcelable. You may choose to make it a global so that you may not need to send it across activities, I would not prefer to do that though.
One UX issue is when user presses up/back it is generally expected that user has cancelled the operation. I as a user expect when I press back it cancels the operation and when I press "apply" it applies the filters. You may need to rethink about the user experience about applying filter on back press.
I have an Android app with multiple activities. The main activity communicates over a network and can launch or dismiss various other activities depending on commands it receives over the network. When an Activity is dismissed I don't want to finish() it, just move it down the stack so it's no longer the top activity. What I really need is a FLAG_ACTIVITY_REORDER_TO_BOTTOM but there is no such thing.
There's an intent flag called FLAG_ACTIVITY_PREVIOUS_IS_TOP and the name implies something like that but I don't understand the description:
"If set and this intent is being used to launch a new activity from an
existing one, the current activity will not be counted as the top
activity for deciding whether the new intent should be delivered to
the top instead of starting a new one. The previous activity will be
used as the top, with the assumption being that the current activity
will finish itself immediately"
Could someone please decode that for me, and if it's not what I want IS there some way to tell an activity to submerge itself below the previous one?
This isn't possible. The activities are stacked and you cant put one back under the other. It sounds like you may want to create a class that extends Android’s android.app.Application.
I found this tutorial online and it looks good. Good luck.
Extending Android's android.app.Application tutorial
You cannot move an activity below a certain activity into the android back Stack. The only way to move a activity in back stack is to open another activity on top of it. You can move an activity on top by creating a single instance of activity using FLAG 'singleTop' in this way your activity will be moved to the top of another activity and only a single instance of activity will be there in stack.
More information about activity back stack and Flags is available here.
Go through this information and all your doubts will get cleared about back stack.
In my app I have situations where I need to get a user back to some activity that preceded(not necessarilly directly) the current one. All of those previous acitivities might need Intent parameters in onCreate.
So, my question is there any easy way to get user back to an activity that might not be the direct previous activity he's been on and is it possible to avoid manual workaround of saving/restoring those previous activities' intent parameters ?
Consider an example: there's a global search-bar that can provide users with suggestions on products; once they hit one of suggested items they get moved on a product-view activity where they can reload this activity with another product - walk through. After a couple of such reloads they might decide to go back to the activity where the search was initiated, but it might not the closest to the current one.
UPD: There also should be a possibility to go back in B activities sequence.
Using startActivityForResult() while loading a new activity and using finish() to close the launched activity on back press can solve your problem.