Archive for category mobility

Building iPhone applications using MonoTouch, part 4: the UISearchDisplayController

Posted by Richard de Zwart in .NET, mobility, technical on November 3rd, 2009

In my previous post I wrote about the Interface Builder and things like outlets. Last night (with the help of some colleagues) I cracked one of the more advanced Classes in the Interface Builder: the UISearchDisplayController.

In the previous version of my application I had a UITableView and a UISearchBar. I hooked them up with some code and it worked fine. But I didn’t get the effect that you see in (for instance) the iPod application. When you scroll up,  uncover the search-bar and start typing, the original view is greyed-out. And when your search gives no result, you get a “No Results” message. Like this:

SearchScreenShot NoResults

For that, you need the UISearchDisplayController. This controller does the work of hooking up a couple of UI-elements for you:

  1. The search bar
  2. The view with the results from the search (called the searchContentsController)
  3. The delegate that handles all the events that come from the search bar and the results view (called the searchResultsDelegate)
  4. The data source that provides the data to search in (called the searchResultsDataSource)
  5. Your original view

When you drag a UISearchDisplayController and drop it at the top of your UITableViewController, all the outlets get connected to your controller automatically. Apparently the Interface Builder thinks that your class can play all these roles. This makes sense when you program Objective-C, since that language is quite capable of inheriting from lots of base classes (as is so eloquently explained in the Cocao With Love blog), but asks some more attention when used from MonoTouch.

Designers should not write code, and vice versa

I will explain again what Interface Builder does for you. Maybe you already know, but I had to get used to it, and have to keep reminding myself that you use it to, well,  build interfaces. Nothing else. Interface Builder provides a clean separation between the GUI and the code, and that’s a good thing, right? Right. I love Design Patterns, and I try to convince as much progammers as possible to take at least notice of them. But iPhone apps are mostly very small apps with one or two screens, being very good in just one or two things. Do I need to implement this whole pattern for my simple app? Well, apparently. And so will you, so let me try to help you find out how to do some of these things.

Steps to follow

Begin with a UITableViewController. Then go to the Library Window, the Objects button and drag-n-drop a UISearchDisplayController just at the top of your UITableViewController.
In the MainWindow you will have the UISearchDisplayController added. Click it there, and then go to the Outlets-tab in the Inspector Window. You will see that all the outlets will be connected to your UITableViewController, except for the searchBar-outlet since that is of course connected to the SearchBar.

When you run your app, you will have the Table-View and the SearchBar above it. When you tap the text-field you will see the desired gray-out and the “No Result” if you enter some text. So far so good.

Providing the view with data

Now we need to hook up some of our own code to the events that the UISearchDisplayController fires. Let’s start with some data in our own TableView to filter on.
As said in one of my first posts, data for a view is delivered by a DataSource. In this case a UITableViewDataSource. So add a class to your project that inherits from UITableViewDataSource. Something like this:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

public class LeesPlankjeDataSource : UITableViewDataSource
{
	private string[] woordjes = new string[] {"aap", "noot", "mies"};
 
	public LeesPlankjeDataSource()
	{
	}
 
	public LeesPlankjeDataSource(string filter)
	{
		woordjes = woordjes.Where(f => f.StartsWith(filter)).ToArray();
	}
 
	public override UITableViewCell GetCell (UITableView tableView, MonoTouch.Foundation.NSIndexPath indexPath)
	{
		UITableViewCell cell = tableView.DequeueReusableCell("plankje");
		if (cell == null)
		{
			cell = new UITableViewCell(UITableViewCellStyle.Default, "plankje");
		}
		cell.TextLabel.Text = woordjes[indexPath.Row];
		return cell;
	}
 
	public override int RowsInSection (UITableView tableview, int section)
	{
		return woordjes.Length;
	}
}

The complete code of this solution is at the bottom of this post.

The class has two constructors. The default constructor (at line 5) is called when initializing our own view, the constructor that takes a string (at line 9) will be called when filtering is started.
The GetCell() and RowsInSection() methods need to be implemented to make your data source work. The implementation is pretty straightforward. The GetCell() method will be called “RowsInSection” times. The call to DequeueReusableCell() is some trick to limit the amount of resources that your iPhone application will use. Just make sure you pass in some string that you reuse a few lines down.

To be able to set this datasource on the table-view we have to have some programmatic access to the view. Well, we did that before in the previous post. Go to Interface Builder, select the AppDelegate in the Library Window, add an outlet and connect it to your UITableViewController. Then you can have code like this in your main.cs:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

public partial class AppDelegate : UIApplicationDelegate
{
	// This method is invoked when the application has loaded its UI and its ready to run
	public override bool FinishedLaunching (UIApplication app, NSDictionary options)
	{
		tableView.DataSource = new LeesPlankjeDataSource();
 
		// If you have defined a view, add it here:
		window.AddSubview (tableView);
		window.MakeKeyAndVisible ();
 
		return true;
	}
 
	// This method is required in iPhoneOS 3.0
	public override void OnActivated (UIApplication application)
	{
	}
}

We simply set the DataSource on the tableView to our own DataSource and then add the tableView to the current window. Run your app and you will have some data in your view!

Building the delegate

Now we need some code to handle the events of the search bar. The most interesting event is “ShouldReloadForSearchString()”.

Add a new class to your project that inherits from UISearchDisplayDelegate. The code should look something like this:

1
2
3
4
5
6
7
8
9
10
11

[MonoTouch.Foundation.Register("LeesPlankjeDelegate")]
public class LeesPlankjeDelegate : UISearchDisplayDelegate
{
	public override bool ShouldReloadForSearchString (UISearchDisplayController controller, string forSearchString)
	{
		Console.WriteLine("In ShouldReloadForSearchString");
		controller.SearchResultsDataSource = new LeesPlankjeDataSource(forSearchString);
		return true;
		//return base.ShouldReloadForSearchtring (controller, forSearchString);
	}
}

The first line is interesting. This is the magic that brings your classes into the Interface Builder. By registering the name of your class it will be added to the XIB (although not visible in the designer file). I’ll show you in a minute what you do with this in Interface Builder.

In the override of the ShouldReloadForSearchtring() I instantiate a new data source using the constructor that accepts a filter string. I set this on the SearchResultsDataSource property of the passed in controller object. As you can see in the code of the LeesPlankjeDataSource it will use a Lambda to filter the fixed array of words.

Hooking up the UISearchDisplayController with your Delegate

The Register-attribute on your delegate class makes it available in Interface Builder. So you go to the Library Window, choose the Objects button, then Controllers folder and then the general NSObject. Something like this: LibraryWindow

Drag it to the MainWindow, select it there, go to the Inspector, choose the Identity tab. Now change the class field to the name of your own class. LeesPlankjeDelegate in my case. Your class will not be listed, but that doesn’t matter. When you hit Enter, you’ll see in the MainWindow that both the class name and the instance name have changed. That is just fine.

Now the next magical thing: you have to connect the default delegate of the UISearchDisplayController to your Delegate class. Here is how: select the UISearchDisplayController in the MainWindow, go to the inspector, select the Outlets tab. The first outlet there is called “delegate” and is connected to your TableView. Now remove that connection by clicking the X. Then connect this delegate to your own Delegate class in the MainWindow.

Save in Interface Builder, go to MonoDevelop, run! Type something in the search and “Lo and behold!” it works!

Ain’t live sweet?

If it doesn’t work, feel free to leave a comment. I’ll see if I can help you.

Download the source code

P.S.

The last step is actually more complex than it should have been. If I make my UISearchDisplayController visible to my AppDelegate by adding an outlet, I can do with just one more line of code in my main.cs:

searchDisplayController.Delegate = new LeesPlankjeDelegate();

That way I go one-way: from Interface Builder to MonoTouch. But I thought it more interesting to go the other way too: from MonoTouch to Interface Builder.

, , , , , ,

21 Comments

Building iPhone applications using MonoTouch, part 3 : The Interface Builder

Posted by Richard de Zwart in .NET, mobility, technical on October 24th, 2009

So it’s about time I tell something about the Interface Builder. As I mentioned in my previous post, I fled from the Interface Builder at first. I’m not a designer, so interfaces are hard for me anyway. But I built some in WinForms and in ASP.NET and I got used to the simple system of putting buttons and fields on forms and then hooking them up with the business-logic with things like click-events.

Interface Builder is different. I think the idea is that you use it for nothing else than your GUI. There is no way to put any code anywhere. All handling of UI-events is done in delegates and the trick is to learn how to link your GUI to your code.

I hope you will have a better insight in how to do this, after this post. I learned the most from one of the videos on MonoTouch.Info, by Code Snack. I literally played it frame by frame, carefully watching what he did and what could be learned from the code on the background. Highly recommended.

So, let us build an interface with a tab-bar-controller. You might know this type of application from the default iPhone clock.

Start a new application in MonoDevelop, an “iPhone MonoTouch Project”. Double-click the MainWindow.xib and Interface Builder will start.
The empty window you get is important because it is the window that is referenced from AppDelegate. You will add all your own stuff to it as subwindows. But for now you can ignore it.

Make sure the Library Window is open (Tools / Library) and selected like this:Library Window

Now drag a Tab Bar Controller and drop it on the MainWindow (that looks a bit like a solution explorer) just below “Window”. The MainWindow will look like this:
Main Window And you get a new window looking like this:
Tab Bar Controller

It is not to hard to change some of the settings of the tab bar. Open Tools/Inspector, select the Attributes tab, make sure you have the right part active in the Tab Bar Window and change the Identifier Pull-down to “Search”. You will get a default Search Tab, including the search image.

Search TabMake the first tab a Search tab and the second a History tab. Now save the project and check the MainWindow.xib.designer.cs. Nothing interesting, except for a mapping to the window called “Window”. If you run your application now, nothing interesting.

That is because we have to add our new view as sub-views to the main window. In code like this:

1
2
3
4
5
6
7
8

public override bool FinishedLaunching (UIApplication app, NSDictionary options)
{
	// If you have defined a view, add it here:
	window.AddSubview (tabBar.View);
	window.MakeKeyAndVisible ();
 
	return true;
}

But we have no variable named “tabBar”. Well, now we’re coming to the very important part where we hook up our GUI to our code. Interface Builder does that with something they call “outlets”. And defining them is one of the weirdest experiences in your live as a programmer.

Ok, since I want to have a reference in my AppDelegate to the Tab Bar Controller, I go back to Inteface Builder, to the Library Window and select the Classes button. All the way on the top, there is a AppDelegate. Select it. Then look on the bottom of the window. Select the Outlets button. Now you see that the AppDelegate has an outlet called “window” which is mapped in the designer file in MonoDevelop to a property of type UIWindow with the name “window”. That is how the code above can reference a “window” object. We need to add an outlet by clicking the + button. Give it a name like “tabBar”. The change the type to UITabBarController.

Now we have a definition of an outlet. But it is not linked to any UI-element yet. So go to the MainWindow and select the AppDelegate there. Now go to the Inspector Window and select the Connections tab. You will see that the connection from the tabBar outlet is still open:
Defined OutletsNow comes the magic. You drag the open circle to the Tab Bar Controller window. While getting there you will notice that only windows that are of the right type (UITabBarController) will be an acceptable drop-target. Now drop it, save it and go to MonoDevelop. In the designer file you will see that a new mapping is made. If you change the code in your main.cs to reflect the code above and run your application, you will have your tab bars!

Ok, so far for this post. Next post I will add a search-bar to the search window on the search tab and try to hook them all up to each other and to the code.

, , , , ,

3 Comments

Building iPhone applications with MonoTouch, part 2 : The AppDelegate

Posted by Richard de Zwart in .NET, mobility, technical on October 20th, 2009

This is part 2 of a series. In the first post I said something about the application structure.

A word about Twitter

I’m going to digress a little before telling you more about iPhone application structure. It is not essential for this series, but it is essential to completely describe the adventure that this first application brought on me.

When I found out about the MonoTouch project, it was in Beta and you had to apply to be admitted. I filled in the form (from my iPhone, just to impress them) and then waited. Nothing for a couple of days. I applied again, just to make sure you know. Nothing. Now I’m a modern guy, so I tweet. I tweeted my frustration (in Dutch), since I really wanted to get started. But hey, what where the 3 followers I have gonna do about that? Then it turned out that this guy Joseph Hill, somehow monitors all tweets in the world for the MonoTouch keyword. Joseph is one of the people in the MonoTouch project that is really a Person. He pulled some strings and presto! I got my Beta-account.

For me, that’s just amazing. I didn’t think that Twitter would give me much more than just the latest tips from Scott Guthrie, but this is a totally new way to bring people together and build communities.

The AppDelegate

When you create a new iPhone solution in MonoDevelop, you get a couple of things:

New Solution

New Solution

  1. A xib file called MainWindow.xib. This is where your user-interface is defined. It is an XML file that can be used by the Interface Builder. Simply double-click it and the Interface Builder opens up. More about the Interface Builder later. MonoDevelop monitors the file and regenerates the accompanying designer file when the XIB is changed by Interface Builder. The designer file maps the elements from the user-interface to the classes in your project.
  2. A main.cs that contains the entry-point to your code. There is  a partial class called AppDelegate that is mapped to the main window of your UI. The most important method there is “FinishedLaunching”. It gets called when the iPhone is done with all the stuff it does when loading an application. The name is a bit misleading. Your application is definitely not done launching, at best it is done loading. All the stuff that you need to do yourself still has to happen. Like loading some files, initializing your controllers, setting-up your model classes.

Loading your application

There’s more to say about the loading process. Since it takes some time before you even get control about what happens, you have the option to show an image while loading. Add a file to your project with the name “Default.png” and set it’s properties to “Build Action – Content” and it will be picked up by the loader. Your supposed (according to the designer guidelines) to use an image of your app in action, but it is often abused to show a company logo or something. Some applications make a screen-shot of themselves when the application closes and save that as Default.png. That way, the next time you start your app it looks like it quickly continues where it left off.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

// The name AppDelegate is referenced in the MainWindow.xib file.
public partial class AppDelegate : UIApplicationDelegate
{
	// This method is invoked when the application has loaded its UI and its ready to run
	public override bool FinishedLaunching (UIApplication app, NSDictionary options)
	{
		Console.WriteLine("Launched");
 
		window.AddSubview(yourViewHere);
		window.MakeKeyAndVisible();
		return true;
	}
 
	// This method is required in iPhoneOS 3.0
	public override void OnActivated (UIApplication application)
	{
	}
}

You should not be doing too much work in this method. The loader times the duration of this method and kills your application when it takes more than 10 seconds. Or so they say, I never tried it out.

You can use this method to build up your user-interface in code. I went that way first. I had such a hard time getting my head around the Interface Builder that I decided to drop it and build my interface by hand. That worked reasonably but caused unexpected crashes. I knew it had something to do with the way I hooked up the different views, so I went back to Interface Builder and tried and tried and tried. I got it, more or less, and my interface (using a tab-bar) is now stable. In the next episode I will dive into the Interface Builder.

, , , , ,

1 Comment

Building iPhone applications using MonoTouch, part 1 : Application Structure

Posted by Richard de Zwart in .NET, mobility, technical on October 18th, 2009

I finished my first iPhone app this weekend. it is not a very impressing application, but it does what I meant it to do and that is to show movies with Dutch Sign Language coming from the website of the Dutch Gebarencentrum.

Developing for the iPhone is exciting. Well, let me re-phrase that: the result of your development for the iPhone is exciting. Mmmmh, let me re-phrase that again: it was a helluva job to build a working iPhone app, and now that I know how to do it I’m excited!

This post is the first of a couple of post I’m planning to do on development of iPhone applications with .NET. There’s simply too much to learn and to tell to do it all in one post. These posts will not only be about programming C#, but also about design principles, open-source programs, Twitter.

Where do you start?

As a .NET developer you have two choices to program for the iPhone:

  1. You learn the Objective-C programming language and the XCode development environment
  2. You use MonoTouch and MonoDevelop

I started out with the first option, since the second was not available at that time. I found it really hard. Using a c-like syntax is not the hardest thing, but I surely wasn’t used to manage my own memory any more! I got really frustrated since I also had to learn all the other things I’ll talk about in a minute.

The solution came in a podcast from DotNetRocks titled “Rory does iPhone” in which Rory Blyth raved against Objective-C. In that podcast there was a short mention of an open-source initiative called MonoTouch.

Mono

That changed my world radically. I went out looking for the Mono guys and found a very mature platform and active community. if you don’t know: Mono is an open-source initiative that brings the .NET framework to a lot of other platforms, especially the Unix-look-a-likes.
I had heard about Mono before, but always thought it was a project that was way behind the development of the .NET framework. You know, like Microsoft is releasing .NET 4.0 and the Mono guys are proud to announce they now support 2.0! Nothing like that! Mono is on .NET 3.5 and preparing for the changes that 4.0 will bring. They also have a very nice IDE called MonoDevelop that looks a lot like Visual Studio.

With MonoDevelop developing applications on the Mac is so straightforward that I never go back to my VMWare installation with XP and VS2008 to do some quick development. If you think about porting your application (since not all that is available on Windows is there on Linux too), install their Mono Migration Analyzer add-in.

And they have MonoTouch. On the surface it is just a wrapper around the CocaoTouch library that you use to program UI’s on the iPhone.

But is is also a very special compiler that pre-JITs (the call it Ahead Of Time compilation) the entire application. Why? The iPhone OS will not allow you to run a virtual machine. Ever wondered why there is no Java or Flash on the iPhone? They cannot run since they depend on their own VM.
MonoTouch circumvents this problem by generating native code, removing all the unused libraries, classes and methods and then building a Mac App package from that.

What do I have to learn?

When you want to build your own application, you will at least have to learn about:

  1. The structure of iPhone apps
  2. The Cocao Touch library
  3. The Interface Builder
  4. Debugging in XCode
  5. Signing your application
  6. How to distribute your application

I will try to tell you about all of these subjects in different installments.

The structure of an iPhone app

It is essential to know what the iPhone expects from you and how the Cocao Touch library supports that. It is probably  way different from the way you worked in C# and WinForms, and it takes some time to get used to finding out where to put your application logic.

The Cocao Touch library is full of Design Patterns, but the most pervasive, noticable, visible and wonderful is The Model View Controller. Below is an image from the Mac Development Library:

MVC

MVC

This post is not about Design Patterns, so I will not explain how the MVC pattern is supposed to work. The image is just their to help me explain which classes you will have to inherit from.

Cocao development is about inheriting from the classes that wire up the application for you. That means that you only have to override the methods that are of interest to you, but it also means that you essentially have no way to stray from this setup. You use the Cocao classes or you end up building your own library.

These are the classes that implement the MVC pattern for the often-used TableView:

  1. UITableView
  2. UITableViewDataSource
  3. UITableViewDelegate
  4. UITableViewController

The UITableView is in the NIB: the file that is generated for you by the Interface Builder and that holds all the GUI elements.

The UITableViewDataSource holds the data that is displayed in the table. It’s methods are called by the Controller to fill the cells in the table (e.g. GetCell, RowsInSection).

The UITableViewDelegate implements methods for dealing with user-actions that are performed on the view, like “RowSelected” when (yep) a row in the table is selected.

The UITableViewController is the man-in-the-middle. It sits between the DataSource and the TableView, and does things like passing the DataSource filtered data when some search criterion is entered and then telling the TableView to reload with this new data. The controller can use your own model to get this filtered data.

So far for this installment. Next time I will show some code and tell about the AppDelegate, the object that functions as the heart of your application since it receives a lot of interesting messages like “Hey you’re app is started!”.

, , , , ,

9 Comments

ScheduledThreadPoolExecutor horribly broken

Posted by Peter Doornbosch in java, mobility, social, technical on March 29th, 2007

A while ago, I considered using the Java 5 ThreadPoolExecutor class for executing
remote calls asynchronuously. The application I was working on needs to perform remote
calls on large numbers of devices, and as remote calls can take quite a while, you
don’t want these remote calls to wait on each other. Moreover, as remote calls might fail,
e.g. due to network problems, a retry mechanism was also needed. The
ScheduledThreadPoolExecutor, a subclass of the ThreadPoolExecutor, additionally allows you to
schedule a task at a certain delay, which offers a simple and elegant solution for the
retry mechanism: when a remote call fails, I only had to re-schedule it with a
proper (increasing) delay and the scheduler would take care of it.

Thanks to the fact the ThreadPoolExecutors provides an abstract method afterExecute()
that is called after execution of the task, I didn’t have to pollute my task
implementation with retry logic, but could clearly separate these concerns. In the
afterExecute() method of the (subclassed) ScheduledThreadPoolExecutor, I could ask the
task whether it had succeeded, and if not I could simply reschedule it. And all this
just in a few lines of code:

void afterExecute(Runnable task, Throwable exception) {
     if ((RemoteCallerTask) task).failed()) {
         super.schedule(task, 10, TimeUnit.SECONDS);
     }
}

When I first tested it, I got a ClassCastExeception. My first guess was that it might
have something to do with different class loaders, but when I ran it in a debugger it
turned out to be something that realy surprised me: the Runnable task that was passed
to this method was not my do-a-remote-call task that I had passed to the executor, but
something of a completely different type
(ScheduledThreadPoolExecutor$ScheduledFutureTask).

Maybe I misinterpreted the documentation? I went back to the ThreadPoolExecutor
javadoc. It talked about “methods that are called before and after execution of each
task”, and the parameter description claimed the Runnable parameter to be “the runnable
that has completed”. This seemed to match my expectations: you execute a Runnable task
and that is what is passed to afterExecute. That would be the only sensible definition
of a after-execution hook, wouldn’t it? As the source code is the best documentation, I
checked the ThreadPoolExecutor source, which confirmed what I was expecting: the task
that is run is passed to the beforeExecute() and afterExecute() hooks.

A little bit of studying on the ScheduledThreadPoolExecutor source revealed why I got a
ClassCastExeception: it wraps the (user supplied) task in a ScheduledFutureTask object
before passing it to the base class (that puts it in the task queue). One of the
reasons why this wrapper is needed, is because the ScheduledThreadPoolExecutor uses a
DelayQueue to store the tasks and elements of this queue must implemented Delayed
(i.e. have a method that returns the delay). This type of queue sorts tasks based on
the delay: shorter delay comes first. When taking an element from the queue, it blocks
until the delay of the first task has passed. Using this type of queue makes the
implementation of the ScheduledThreadPoolExecutor quite simple: it wraps the task in a
wrapper that implements getDelay() and puts these wrappers in the queue.

Although I can appreciate the beauty of using a DelayQueue in combination with a normal
ThreadPoolExecutor, I don’t think it is the right solution. The point is that it breaks
one of the fundamental principles of OO programming and that is that derived classes
should respect the contract defined by the base class(es) (and or interfaces). The contract
that the base class ThreadPoolExecutor defines, is that it will call the hook methods
with your task as a parameter. ScheduledThreadPoolExecutor breaks this contract, as it
does not adhere to what its base class has promised.

This break of contract shouldn’t be taken lightly. It makes code that uses these
executors fragile: if for some reason someone decides to use the other class as
implementation of the general executor service, existing code might break. Put
differently: in order to make your code robust, you would to have to take into account
which executor implementation was chosen, at different points in your code. This
violates principles of encapsulation and abstraction: code should never depend on
implementation types, only on interfaces.

I was pretty disappointed that even in the concurrency API, such fundamental
mistakes can be found. Moreover, it appeared this is not the only break of contract, and that fixing this properly doesn’t seem to have any priority, but more on that later.

,

1 Comment