In an instance of an UIViewController with ARC best practice would be to "release" any retained subviews of the main view (e.g. self.myOutlet = nil) in - viewDidUnload, and my guess is self.view = nil isn't strictly necessary at this point.
What to do if an additional retained property is defined and assigned as...
self.anotherProperty = self.view;
or
_anotherProperty = self.view; // assuming "_anotherProperty" is the ivar
... is self.anotherProperty = nil necessary then?
Nothing in viewDidUnload (I'm assuming we are talking about when running on pre-iOS 6, as views are no longer unloaded in iOS 6, and thus viewDidUnload is no longer called) is "necessary"; it's only an optional opportunity to unload stuff.
But if you wanted to unload stuff properly following the convention for viewDidUnload, should you nil your property? Yes. 1) If not, your property will still retain the view object, preventing it from being deallocated, which is the point of unloading the view. 2) The view has been unloaded, which means self.view is set to nil, and will be set to a newly loaded view when we decide to load the view. If you want your property to always point to self.view, then you need to follow suit and set it to nil on unload and set it to the loaded view in viewDidLoad.
Related
When cloning a button at runtime using Instantiate(), which contained listeners on it's onClick event, the listeners are not present in the clone.
The behaviour can be tested by having a Canvas with a button and this script attached:
void Start () {
var button = transform.GetChild (0);
button.GetComponent<Button> ().onClick.AddListener (new UnityAction(() => Debug.Log("Event triggered!")));
var button2 = Instantiate (button);
button2.SetParent (transform);
}
The cloned button will not print anything to the console when clicked.
Is there a way to clone a GameObject so that it retains event listeners?
Runtime listeners are not persistent and then not serialized. As a result they are not passed on when you clone the button.
Either you'd have to add the method to a script and attach the script to your prefab for it to be serialized along or assign it by code like you do for the first one.
Instantiation. When you call Instantiate() on either a prefab, or a
gameobject that lives in the scene, or on anything else for that
matter (everything that derives from UnityEngine.Object can be
serialized), we serialize the object, then create a new object, and
then we “deserialize” the data onto the new object. (We then run the
same serialization code again in a different variant, where we use it
to report which other UnityEngine.Object’s are being referenced. We
then check for all referenced UnityEngine.Object’s if they are part
of the data being Instantiated(). If the reference is pointing to
something “external” (like a texture) we keep that reference as it
is, if it is pointing to something “internal” (like a child
gameobject), we patch the reference to the corresponding copy).
http://blogs.unity3d.com/2014/06/24/serialization-in-unity/
As for AddListener
This function adds a "non persistent" delegate, which means it will not show up in the inspector, and will be forgotten when you exit play mode in the editor. These differ from "persistent" listeners which you can add during edit-time in the inspector, and which persist between edit and play mode.
https://docs.unity3d.com/ScriptReference/Events.UnityEvent.AddListener.html
For those reasons, I assume the non-persistent is not serialized and then not passed to the cloned object.
I have a browser view that on its call method has something like this:
def __call__(self):
context = aq_inner(self.context)
parent = aq_parent(context).
...
Putting a pdb at the beginning and playing with it seems that, for Dexterity there is no need to use it, is that right?
ipdb> self.context, id(self.context), self.context.__class__
(<Container at /plone/ausgaben>, 4651890160, <class 'plone.dexterity.content.Container'>)
ipdb> aq_inner(self.context), id(aq_inner(self.context)), aq_inner(self.context).__class__
(<Container at /plone/ausgaben>, 4651890160, <class 'plone.dexterity.content.Container'>)
So the result is the same using aq_inner or not.
So the question is: does Dexterity (as self.context and in our project actually everything is Dexterity based) prevent us from having to wrap everything with aq_inner and aq_parent and so on, and instead use directly the objects or __parent__ pointers?
Like AT contenttypes, DX contenttypes are also aq-wrapped. So you're going to face the same behaviour (issues :-)) as with AT.
As sdupton said in his aq_parent(instance) == instance.__parent__. The parent pointer ist still implemented thru acquisition.
But there is a small differences to AT.
If you create a new DX obj the following happens:
createContent will be called which creates the DX obj - At this point the content is not yet aq-wrapped. So if you subscripe the ObjectCreatedEvent you gonna have a not aq-wrapped obj.
addContentToContainer will be called, which adds the created DX content to the container. In container._setObject the ObjectAddedEvent will be fired. If you subscribe this event you will have a aq-wrapped dx content.
This is different in AT, of course other events are fired for this case, but the AT content is always aq-wrapped (also in the factory, while adding a new AT obj)
Please let me know, if I misunderstood something.
My problem is with an iOS 6 tabbed application. My work-in-progress has 5 tabs and several tabs are gateways to other view controllers. Most pages need access to a Model object, which contains data stored as arrays, strings, and so on. A bare-bones model is populated at runtime and the user can add to it throughout the application lifespan. For example, the code listed below is from my AppDelegate file , where it is passing a pointer to the bare-bones Model to the Project View Controller. This works fine: the tab application uses the navigation controller array stack; because I know the Project page is at index 2, I can send the model to it.
My problem has to do with the sub views of the Project page. For example, as a sub view to the Project page there is (or should be) a File_IO page where the user handles file operations. The File_IO page also needs access to the Model object. But when I try to send the Model pointer from the Project page to the File_IO page, the technique I used previously (from the AppDelegate to the Project) does not work. I point to an empty Model in the FileIO ViewController.
Example code: this is in the AppDelegate, and it works fine: the bare-bones Model in the Project ViewController is populated with the data.
//To the Project View Controller...
UINavigationController *project_NavigationController =
[[tabBarController viewControllers] objectAtIndex:2];
Project_ViewController *project_ViewController =
[[project_NavigationController viewControllers] objectAtIndex:0];
//This hides the navigation bar on the project page but it also removes the bar on any subsequent pages! They need to be programmmatically reset in ViewDidLoad.
[project_NavigationController setNavigationBarHidden:YES animated:YES];
project_ViewController.currentGeoModel = newGeoModel;
Now, my Project_ViewController is embedded in a NavigationController and has 4 child ViweControllers. One is named FileIO_ViewController. I want to pass the currentModel from the Project_ViewController to the FileIO_ViewController. Under the - (void)viewDidLoad method I have tried a number of things, which do not work. For example,
UINavigationController *project_NavigationController = (UINavigationController *) self.presentedViewController;
FileIO_ViewController *fileIO_ViewController = [[project_NavigationController viewControllers] objectAtIndex:1];
fileIO_ViewController.currentModel = currentModel;
compiles but when I try to access currentModel inside the FileIO_ViewController methods, it is empty.
If anyone can take the time to help I would be very appreciative. The best answer for me would be in the form of an explicit code example showing how to pass the pointer to an object like my Model from a ViewController to another ViewController where you do not explicitly know where in the stack the child VC lies. (In my example I used Index 1 but I do not actually know at which Index the FileIO_ViewController lives as I have three other ViewControllers under the Project_ViewController. I've tried several integers with no success.)
If you do answer this, please consider me a New Guy when it comes to iOS 6 and objective C -- climbing Mount Apple has been a long haul and I isn't anywhere near the top yet!
Thanks
Tim Redfield
Norway
If you have a shared single model for your app, you shouldn't proactively pass the pointer around, you should make the model available from a single location and leave it to individual objects to access this same model when they need to. A good location for your model pointer is in your Application delegate.
In the appDelegate's .h file, declare a property for your model:
//appDelegate.h
#property (nonatomic, strong) MyAppModel* appModel;
After you instantiate your model in the appDelegate, just assign it to the property:
//appDelegate.m
- (BOOL)application:(UIApplication *)application
didFinishLaunchingWithOptions:(NSDictionary *)launchOptions {
self.appModel = [[MyAppModel alloc] init];
//set up bare-bones appModel here
return YES;
}
Then you can access this property from any viewController that needs model data thus:
#import appDelegate.h;
AppDelegate* appDelegate = [UIApplication sharedApplication] delegate;
MyModel* model = appDelegate.model
(Better still, make the model object into it's own Singleton object, but using the appDelegate will get you started)
If you need to pass models around, then you need to take care that you are passing them to the right object. This is where your existing code is breaking. For example you are making many assumptions about the structure of a NavigationController stack. But whenever you move back down a stack by popping a controller off the top, you lose that top controller instance completely. When you 'return' to that controller by going forwards on the stack, in fact a new instance is created (unless you have taken care to keep a pointer hanging around and make sure to push to that pointer). If you need more help on this aspect, you will need to describe exactly the layout of your app, and how you are creating your viewControllers, navigationController stack, etc.
update (following your comments)
If your changes to the model are valid throughout the app, then I don't see why you can't just access the model when you need to from wherever you happen to be in the app via appDelegate.model. If you have concurrent versions of the model, you could look at making a singleton data manager object which could handle the details of storing an array of models (or a model history) and providing the correct model as per request. The principle is the same - rather than proactively passing model objects into your viewControllers, let them request data from a centralised source as they need it.
Regarding your description "Now, my Project_ViewController is embedded in a NavigationController and has 4 child ViewControllers.", I don't think you have quite grasped the distinction between Classes, Storyboard scenes, and instances.
You have this arrangement in a part of your storyboard:
UINavigationController
| push push push
|->UIViewController1 -----> UIViewController2 -----> UIViewController3 ----->
segue segue segue
You talk about passing data directly from VC1 (say) to VC3 by accessing the NavController's stack.
What you need to consider is that the storyboard describes a template showing how instances will interrelate when they are instantiated. They are not instantiated just because the storyboard is loaded. When you have arrived at VC1, the ability to segue to VC2 and VC3 is laid out before you in the template, but neither VC2 nor VC3 - as instances - exist until you initiate the segue. The segue process instantiates it's destinationViewController Therefore it makes no sense to pass data from VC1 directly into VC3. When you are at VC1, the navController's stack only contains one item (a VC1 instance); when you segue to VC2, it then contains instances of VC1 and VC2, and it is only when you actually segue to VC3 that the instance is created and placed in the stack.
Stepping through your code:
UINavigationController *project_NavigationController =
(UINavigationController *) self.presentedViewController;
the presentedViewController property works with modal segues, where you have a presenting, and a presented, view Controller. NavControllers on the other hand work with push segues (as they push child viewControllers onto their viewControllers stack, which is where you can obtain pointers to them from).
In this context, self.presentedViewController is nil, which you are assigning to a new variable. The code does nothing, but the compiler won't complain.
FileIO_ViewController *fileIO_ViewController =
[[project_NavigationController viewControllers] objectAtIndex:1];
project_NavigationControlle is nil, so it's properties are all nil. fileIO_ViewController gets nil.
Likewise in the last line you are sending a message to nil, which is not an error, but faintly redundant.
I have developed my first flex mobile application which is of TabbedViewNavigatorApplication. Application is working fine but when I test the application in "profile handler", memory usage goes on increasing as I navigate through the application. When I came to know that, I have to remove all the added eventlisteners and I have to nullify the objects which are no longer needed. When I switch between tabs , tabs are initialising again and again.
I dont know where can I remove the eventlisteners. I mean, I have written functions for each eventlisteners . Do I need to remove eventlistener when control goes to the function definition.
I have written sample code
var more:Image = new Image();
more.width = 70;
more.height=29;
more.x=10;
more.y=276;
more.source = "Assets/more button.png";
more.addEventListener(MouseEvent.CLICK, MORE_clickHandler);
mainGroup.addElement(more);
private function MORE_clickHandler(e:MouseEvent):void {
// Do I need to remove the eventlistener here
}
Also , do I need to explicitly nullify the object of Image class which I created or garbage collector will handle it. If I need to explicitly nullify it, where to do this.
Thanks
Garbage collection is an important part of any language, especially on mobile. Since mobile devices are a lot more limited than say our desktop counterparts, you need to be very careful what is being created/stored to memory. My motto is, if you don't see it, you shouldn't keep it. You can destroy views but keep their state using a view model.
To remove a view, you need to first remove it from the display list (removeElement(yourObject)), remove all event listeners, and nullify any referencing variable. If any variable still has a reference to it, it won't get garbage collected.
I recommend you read up a bit more on garbage collection as well as some neat tricks like pooling and virtualization (item renderers in a list).
You can setup an event listerener with a weak reference.
This implies that when the only reference to your object is the listener, the object itself can still be garbage collected and the listener will not keep it in memory.
The following will do the trick :
more.addEventListener(MouseEvent.CLICK, MORE_clickHandler,false, 0, true);
Another option would be to subclass the image class and let it implement an IDisposable interface, which would force you to implement a dispose() method.
Some handy resources:
http://www.intriguemedia.net/2007/09/24/when-to-use-weak-references
http://gskinner.com/blog/archives/2006/07/as3_weakly_refe.html
cheers
i have a Flex tree control and im trying to select a tree node 3 levels down right after the dataProvider is assigned with a collection object like the following.
basically treeItem1, treeItem2, treeItem3 are the nodes in the tree and treeitem3 is a child of treeItem2 which is a child of treeItem1. Assume these treeItem(1,2,3) are referenced correctly from the collection items.
my problem is that if i wait for the whole component to load completely then select the nodes, it open/select/scrolltoIndex correctly. However, if i were to select the node right after the dataProvider is assigned, then it doesn't even open or select (basically the this.treeService.selectedItem is always null).
can anyone point out what i did wrong? is there anything needs to happen after the dataProvider is assigned?
thanks
this.treeService.dataProvider = oPricingHelper.getCurrentPricingSercicesTreeSource();
this.treeService.expandItem(treeItem1, true);
this.treeService.expandItem(treeItem2, true);
this.treeService.selectedItem = treeItem3;
this.treeService.scrollToIndex(this.treeService.selectedIndex);
I have used the updateComplete event to know when a component (such as a DataGroup or List) has completed rendering after performing a simple task (such as updating the dataProvider reference). Of course, you have to be careful and remove listening to updateComplete because it can run a lot, unless you have a need for it to run.
Something like:
//...some function...
this.treeService.addEventListener(FlexEvent.UPDATE_COMPLETE, onTreeUpdateComplete);
this.treeService.dataProvider = oPricingHelper.getCurrentPricingSercicesTreeSource();
//...rest of some function...
private function onTreeUpdateComplete(event:FlexEvent):void {
this.treeService.removeEventListener(FlexEvent.UPDATE_COMPLETE, onTreeUpdateComplete);
this.treeService.expandItem(treeItem1, true);
this.treeService.expandItem(treeItem2, true);
this.treeService.selectedItem = treeItem3;
this.treeService.scrollToIndex(this.treeService.selectedIndex);
}
I'm not positive your experiencing the same issue but I seem to have the same type of problem with using the advanced data grid, it appears in these cases where the dataprovider is acceptable as multiple types, the components do some extra work in the background to wrap things up into something Hierarchical (HierarchicalData or HierarchicalCollectionView) and in doing so the dataprovider setter call is not synchronous (so it will return before actually having assigned the internal property storing the dataprovider). I've used callLater in this case with moderate success, callLater is generally a bad practice but basically adds a function to a list of functions to call once background processing is done, so this is assuming that something in the dataprovider setter called UIComponent.suspendBackgroundProcessing() and that it will subsequently call UIComponent.resumeBackgroundProcessing() and then it will execute the list of functions added by using callLater. Alternatively you could use setTimeout(someFunction,1000).
These are both "hacks" the real solution is to dig into the framework code and see what it's really doing when you tell it to set the dataprovider. Wherever you see that it actually has set the dataprovider you could extend that class and dispatch an event that you could listen for to run the function to do the selections after this point.
If anyone has a better solution please by all means correct me (I would love to have a better answer than this)