I've built an iOS 5 iPad app which makes use of a second screen. We have an admin view (on the iPad) and an external view through an HDMI enabled TV connected via the Apple DVI adapter. Both the iPad view and the TV view get the same data updates from a service call which is made every few seconds. We then present the data received as a series of charts; the charted data is presented very differently for the TV and iPad views - but the core dictionary of data is the same. I'm wondering about an elegant way to architect this solution. At the moment I have one of the view controllers (the admin iPad VC) doing the service calls using GCD and then dispatching NSNotifications which update the data (charts) properties on the other (TV) view controller. I'm considering moving the service calls away from the VC and creating a singleton which is initialized in the app controller. I then (somehow) set the two VCs as delegates and they get updated using a simple protocol. I'm not entirely sure if this is a good approach or if I should consider something else? Can I even set both VCs as the delegates of another class or is it typically only one delegate per class instance?
Thanks for any input.
Ben
Why not abstract the chart data into its own model class, which you can share in both view controllers? The model class can be responsible for fetching the new data. To make the controllers aware of updates, they can either use KVO on the model object, or they can observe notifications sent from the model object when an update occurs, or you can have an array of delegates for the model object and each view controller can be a delegate.
There doesn't seem to be any compelling reason to make it a singleton, although you can if you really want.
Related
Lately I have been playing around with the bluetooth framework and grew a strong enough knowledge to start building an application. The only problem is that all the examples I found and all the practice I have made consist in putting the core bluetooth core code inside the same file as the UIView with which the user is interacting.
I would like my future application to have multiple views in which the BLE scan occurs on the background. I initially thought about creating an object with a name similar to bleDeviceFinder and pass this object through each view. However, after thinking about it I realised that if I want something to happen in the current view I need the function didDiscoverPeripheral to have direct access to the UIView objects which it is supposed to affect.
I know it is probably a stupid question, what would be the best way to do so? I was thinking maybe to set and alert and subscribe every view to that alert; is this a good solution?
A quasi singleton BTLEManager that you pass around in the app. It sends NSNotifications for events like discovery, and your ViewControllers observe these notifications. The truth (i.e. list of discovered devices) stays in BTLEManager. Once a viewController has received such a notification it asks the BTLEManager for the list of current devices and then the viewController changes your views accordingly. The Views should never talk to the BTLEManager directly.
That's how I would do it.
I'm an iOS developer just starting to develop for OS X. In iOS, I always follow the MVC pattern, and put appropriate code in UIViewController subclasses. The App Delegate simply initializes the main view controller, and that's it. After that the view controllers handle all logic.
But in OS X, I seem to put everything in the app delegate. My app delegate class is now about 300 lines long and I am wondering how I can follow the same pattern I did in iOS. There are no view controllers!
What I mean is, what is the common accepted way to organize code for Mac Application?
You can create your own controller classes, and put an instantiation of these controller classes between the app delegate or model objects, and your view objects. Just because there is no pre-built view controller class, does not mean that you can't make your own similar partition, by creating custom controller classes of your own, as needed.
I wonder what the optimal way is to respond to notifications sent out by the notification centers.
Here is an example:
I have a model which receives updates from the server.
Whenever new information is received a notification is generated and posted though the NSNotificationCenter.
There is a view controller with lots of (partially nested) subviews; depending on the type of information received I have to update a particular subview.
For me, there are currently two solutions:
The view controller becomes observer and tells a particular view to update based on the notification name. [subviewx pleaseUpdate];
Each view registers an a observer and depending on the notification name.
The downside of 1 is that the vc has to deal with all notifications although he is not really affected.
Is there any proposed way to do this? Should the responsible view controller deal with all notifications or is it ok for a UILabel, for instance, to become an observer and be somewhat independent.
Thanks for your opinion!
An interesting question - technically, both approaches produce the same result.
However, personally I'd lean towards keeping your notification handling in the view controller, because that's closer to the model-view-controller (MVC) pattern in iOS.
The other advantage of having your notification in the view controller is that you may want to reuse your views elsewhere in your app, and you don't want adverse side effects happening when views start responding to notifications they weren't intended to receive. Collating all your notifications in the view controller will also make handling them much easier - don't forget you need to remove your notification observers when you're done with the view, and having all your removeObserver statements in one place is arguably far better than spread across multiple classes.
I'm coding my first app in XCode 4.2 supporting iOS 4.x through the current releases. To support iOS 4.0, I'm not using Storyboard feature and also using unsafe_unretained for weak references. I have AppDelegate files (.h and .m) along with several view controllers with UITabBarController. In my first view controller, in the -viewDidLoad method, I initialize two NSDictionaries and also start a timer with 1 sec interval. In the selector method, I have to pickup a random number between 0 and 7 to pick a corresponding value in both the dictionaries. The dictionaries are used only in the first view controller and not anywhere.
My first question is
where do I load those two dictionaries - in the AppDelegate -didFinishLaunchingWithOptions: method or in the first view controller's -viewDidLoad method?
I also wanted to support iPad. If that's the case, do I create a common class library to support iPhone/iPod/iPad?. If that is the recommended way, can I move the common functionality to the AppDelegate .m file instead?
Please advise.
You can move your common data and business logic into a separate set of model classes outside of the UI layers and appdelegates. This is one of the main benefits of the MVC pattern - by separating and making a clear distinction, it's easy to have separate view layers (one for phone and one for iPad).
That means all the data (dictionaries), logic with your random numbers and timers would be encapsulated and shared. That also allows you to cleanly unit test the majority model and logic programmatically. It also means you can make substantial changes to your algorithms and minimize the code churn.
When the timer goes off it can either post a notification or you can have a delegate pattern where it does a call back.
Related Post: Delegates vs. events in Cocoa Touch
If you do a shared model, one option is to use a singleton pattern where you access the model like:
[MyModel sharedInstance];
You should keep your code and data together if possible, which means that if you only access the dictionaries in your view controller then you should initialize them in the view controller's viewDidLoad.
I recommend keeping stuff out of the app delegate if possible or else you'll end up with a weird monster class that does too much stuff for which it shouldn't be responsible. If necessary, create one or more classes that manage common data (for example using the singleton pattern).
Whether you can/should use a common file for both iPhone and iPad depends on a number of factors. The main factor is: how different are the UIs? If they are very similar then use one class for both. You can also create a base class with common functionality and sub classes for iPhone and iPad which implement the necessary differences.
I'm relatively new to Objective-C and Cocoa... I've been trying to understand how to correctly implement the MVC pattern in Cocoa/Cocoa Touch for a long time now... I understand the idea behind it; it makes complete sense conceptually: a model holds the data, a view is what that the user sees and can interact with, and the controller acts as the bridge between the two. View can't talk to the model, model can't talk to the view. Got it.
What doesn't make sense to me is how to use MVC efficiently… if the user can only interact with the view, and does something to interact with it (i.e. for an iPhone app, the user clicks/drags within a subclass of UIView, triggering the "touchesBegan" and "touchesMoved" methods, etc.), how does the view communicate these events to the controller?
I've looked at countless examples and forums online, but have yet to find a simplified all-purpose way of achieving this… I know how to communicate with a controller through buttons, sliders, and other things that you can connect to an outlet, but for things that don't have a target-action mechanism, what's the best way to do it?
Thanks in advance for any suggestions regarding what to do, or where to look.
The standard way in Cocoa to do this is the delegate pattern (cf. UITableViewDelegate). Your view class would declare a delegate protocol and the controller sets itself as the view's delegate. The view then calls one of the delegate methods you defined whenever it wants to communicate something to the controller.
An alternative would be to implement the target-action mechanism for your view yourself. You get this more or less for free if you subclass from UIControl (just call sendActionsForControlEvents:) but it is quite easy to implement a system that works the same way for any custom class.
(Edit: I suppose a third way is to have the controller observe properties of the view (with KVO). This wouldn't work well to communicate touch events but it is a feasible way if you want to notify the controller about a state change or something like that.)