A few weeks ago, just two days before Windows 7 become generally available, Visual Studio 2010 hit its own major milestone with the release its second Community Technical Preview of Visual Studio, known as Visual Studio 2010 Beta 2. To me, it is always exciting to see how the different tools and frameworks evolve and add new features.  It seems that with every release the products get bigger and better, offering an even


greater number of programming  languages, and addressing an ever growing number of areas of development such as Web, client, mobile, parallel, consoles, and devices.

Despite being a “beta” product, it is much easier to work with Visual Studio 2010 than with VS 2008. It is much easier to control your solutions and, even more importantly, much easier to write and document code. The user interface is much improved; it uses the Windows Presentation Foundation (WPF) to reduce clutter and visual complexity, and modernizes the interface by removing outdated 3D bevels. Using WPF enables us to help developers focus on content areas by opening up negative space between windows and drawing attention to the current focus with a dominant accent color and a distinctive background. There are also some cool, small, and simple perks like the ability to control the size of text. You can also drag a single window from the main Visual studio application to a second monitor (just like that) and with that you have multi-monitor support via the VS client application.

Improvements to the IntelliSense allow it finally to work well for C++ projects. And let’s not forget the new debugger window that supports parallel computing debugging and lets you view your parallel stacks.

There is even a new language, F# (F stands for Functional programming), and numerous upgrades to C#, like support for dynamic keywords. Dynamic objects' operations are resolved at runtime (check out a good post about this by Scott Hanselman). There is also support for the next version of the C++ language specification, C++X0, like Lambda Expressions. Speaking of C++, we've built the C++ solutions using MSBuild, which should make everyone happy.

As always, backward compatibility is super critical, and it is important to mention that Visual Studio 2010 supports multi-targeting. Visual Studio 2010 can target .NET 2.0 through .NET 4.0 on a per-project basis, which means you can work with your older project on the new VS 2010 and enjoy all the goodies mentioned above (and many more).

But I want to focus this post on using Visual Studio 2010 to program Windows 7. There are quite few technologies and features in Visual Studio 2010 to help you write better applications targeting the specific features of Windows 7. Below are just a few of the Visual Studio 2010 features that we’ll write more about really soon.

.NET 4 and Windows 7

Visual Studio 2010 brings a complete new CLR version – version 4. This is not just an incremental upgrade on top of CLR 2 (.NET Framework 2). This enables new language enhancements like the dynamic keyword. And the new WPF brings support in a few other areas like shell and Taskbar integration, and multitouch.

WPF & Taskbar Integration

As you know, you program jump lists using the JumpList class. This exposes several methods and properties that manipulate the exposed jump lists for the application. It also has an attached property that you can apply to your application class to create, modify, and remove jump list items. If you work with specific files, you can use the JumpList.AddToRecentCategory method to add that file to the recently used file list managed by the shell.

There are two types of jump lists – tasks and items; you work with each using a JumpTask, or a JumpPath respectively. You can work with these in XAML, code-behind, or a combination of the two. The following code snippet shows a simple integration of tasks into a jump list.

      <JumpTask ApplicationPath="notepad.exe" 
                         CustomCategory="External Tools" 
                Description="Take Notes" 
                Title="Start Notepad" 
                IconResourceIndex="0" />

      <JumpTask ApplicationPath="calc.exe"
                         CustomCategory="External Tools" 
                         Description="Perform some calculations" 
                        Title="Start Calculator"
                        IconResourceIndex="0" />

In a similar way, you can use XAML to add Thumbnail Toolbar buttons as shown by the following code snippet.

   <ThumbButtonInfo DismissWhenClicked="True" 
                    Command="{Binding BookItinerary}"
                    Description="Book the itinerary now" />

WPF Common File Dialog Supports Libraries (Finally!)

For some strange reason, WPF 3 and 3.5 Common File Dialog (CFD) didn’t support the updated version of the CFD introduced with Windows Vista. The CFD was upgraded in Windows 7 to support libraries and provide a better user experience. It now allows seamless search integration as well as some advanced user functionality. With WPF 4, applications enjoy the power of the “new” CFD directly from WPF, and don’t need to import CFD from the WinForm namespace (which was the only way to show the updated CFD from WPF 3 and WPF 3.5).

WPF Supports Multitouch

WPF 4.0 introduces multitouch support directly into the WPF API– with no need to interop to a native service. These new features are only available on Windows 7 and are automatically deprecated when running under older operating systems, so you don’t have to detect the operating system yourself. WPF 4.0 adds a new manipulation API to the UIElement base class. This new support allows developers to track multiple touches and generate both cumulative and individual manipulations across the touches. Basically, this enables you to transform your object on the X and Y coordinates, rotation, and scale.

WPF will supply these manipulation events if the IsManipulationEnabled property on the element is set to true. It defaults to false, so you will need to turn on this property for each element where you want to manage manipulations. This is as simple as adding IsManipulationEnabled=true to your XAML as shown in the following code snippet.

<Border Margin="10,5" 

Optionally, you can also hook the ManipulationStarting and ManipulationCompleted events to provide code behind the implementation of these events.

WPF 4 also supports low-level touch messages, or raw touch input. You can interact with the raw touch events on any UIElement using TouchDown, TouchMove, and TouchUp events, all of which have preview event versions. This can be useful if you are trying to track multiple touches that are not manipulating the same object, or if you want to provide different behavior for touches and the mouse. We’ll soon write more about Windows 7 multitouch in general and WPF specifically.

MFC Updates

In Visual Studio 2010, C++ and MFC received a healthy dosage of “coolness” factor, adding useful features such as IntelliSense enhancements and C++0x features. The MFC Library received a major upgrade, especially in regard to the Taskbar, Multitouch, and Restart and Recovery:


The MFC Taskbar provides all the functionality that the native taskbar COM API provides. There is nothing that the one can do that the other cannot. The MFC simply wraps the Win32 APIs (as it always does) into a more “MFC-like” API that corresponds to the MFC Framework programming style guidelines. For example, the following code snippet sets the overlay icon.

CMainFrame* mainFrm = 
if (mainFrm)

First you need to obtain a handle (a pointer) to the application's main window (the top-level window), which corresponds to Win32 HWND. Then, simply call the SetTaskbarOverlayIcon passing HICON and a string that provides an alt text version of the information conveyed by the overlay to meet accessibility requirements. Simple, right?

In MFC, the CFrameWnd class provides the functionality of a Windows single document interface (SDI), overlapped, or pop-up frame window. With the new MFC, this class was updated and now supports Taskbar functionality such as icon overlay, progress bar, jump lists, and thumbnails.

In MFC, Taskbar thumbnail preview support is built in, so the Taskbar thumbnails will show any rendering within the views. Therefore, other than implementing your own View drawing, you need not provide any explicit code to update those Thumbnails.

To enable Taskbar Thumbnails in an MFC application while using the MFC application wizard, all the user needs to do is select the “Multiple documents” application type with the option “Tabbed documents” enabled. When the application runs, MFC will take a snapshot of each view and send it to the Taskbar APIs to display as thumbnails.


And the output could like this:



In Visual Studio 2010, MFC also supports multitouch. By default, on a touch-enabled device (such as touch screen), Windows 7 sends gesture touch messages to any application; that is, by default Windows 7 sends WM_GESTURE messages to the target windows. All that MFC is doing is mapping these messages to its own message handlers. MFC provides a number of message handler overrides that can receive each of the gesture types, and each returns a Boolean value. If a gesture input is processed by the application, the corresponding override should return TRUE; otherwise, it returns FALSE. Therefore if you wish to handle the zoom gestures, all you need to do is implement the relevant handler. Here is the list of supported handlers.

// Gesture handlers
virtual BOOL OnGestureZoom(CPoint ptCenter, long lDelta);
virtual BOOL OnGesturePan(CPoint ptFrom, CPoint ptTo);
virtual BOOL OnGestureRotate(CPoint ptCenter, double dblAngle);
virtual BOOL OnGesturePressAndTap(CPoint ptFirstFinger, long lDelta);
virtual BOOL OnGestureTwoFingerTap(CPoint ptCenter);


Similarly, you can register to receive raw touch messages and the default gesture messages. In Windows 7, gestures messages and raw touch are mutually exclusive. If you register to receive the raw touch messages for a particular window, that window will stop receiving gestures messages. If you opt-in to handle raw touch messages, you need to implement the following handler:

virtual BOOL OnTouchInput(
                        CPoint pt, 
                        int nInputNumber, 
                        int nInputsCount, 
                        PTOUCHINPUT pInput);

MFC makes your life easier by providing a lot of the information per each touch point, for example, the client coordinates for the actual point where the touch-enabled device has been touched. MFC also provides the ID of the touch point, that is, the first, second, or third finger, as well as the total count of current touches.

Restart and Recovery (Restart Manager)

In Visual Studio 2010, MFC also provides native support of the Restart Manager. Restart Manager is a feature introduced by the Microsoft Windows Vista operating system. It can help applications maintain their data when an update needs to shutdown the application or when an unexpected software error or crash occurs. Instead of shutting down abnormally, Restart Manager enables an application to perform an application save before it is terminated. Furthermore, it can re-invoke the application, enabling it to restore its state from before the shutdown or crash.

For new MFC applications, you can get the application restart and recovery feature for free by using the MFC Application Wizard as you can see from the following image:


All configurable parts of the restart manager API are exposed to the user through virtual members that can be over-ridden. Needless to say, you can expect some more blogging about this feature.

.NET 4 and Location

.NET 4 has a new Device namespace that supports the Windows 7 Location API (part of the Windows 7 Sensor and Location. The System.Device.Location namespace allows application developers to access the user's location easily using a single API. Location information may come from multiple providers, such as GPS, Wi-Fi triangulation, and cell phone tower triangulation. The System.Device.Location classes provide a single API to encapsulate the multiple location providers on a computer and support seamless prioritization and transitioning between them. An application developer using this API does not need to know which location-sensing technologies are available on a particular computer and is freed from the burden of tailoring an application to a specific hardware configuration.

To begin accessing location, you need to create a GeoLocationProvider. This object is the main “location manager” object through which you can register for LocationChange notifications and synchronously read the latest location information. Next you need to call Start to start the acquisition of data from the current location provider. You can check the Status property to determine if data is available. If data is available, you can get the location once from the Location property, or receive continuous location updates using the LocationChanged event. The following code snippet is a VERY simple code sample showing how to retrieve the current GeoCoordinates (latitude, longitude).

GeoLocationProvider provider = new GeoLocationProvider();
GeoCoordinate coordinate = provider.Location.Coordinate;
if (coordinate != GeoCoordinate.Unknown)
  //Business logic here

Unfortunately, .NET 4 supports only the Location API and not the full Sensor and Location Platform – meaning that the .NET location implementation is still missing the Sensor piece. Use the Windows API Code Pack to access sensor from managed code.

Parallel Computing and Windows 7 Multi-Core

Parallel programming in Visual Studio 2010 has many aspects, for example, Parallel LINQ and other .NET enhancements for supporting parallel computing, including statements like Parallel.For that use System.Threading.Tasks.Task. C++ developers will be happy to learn that the Task concepts also exist in C++ Version 10, which ships with VS 2010. For native code, Concurrency Runtime (ConcRT) has implicit knowledge of Win7 processor groups and will schedule work on up to 256 cores; ConcRT also takes advantage of User Mode Scheduling of threads. Therefore, any workload that sits on top of ConcRT immediately benefits. In other words, because both the Parellel Pattern Library (PPL) and Asynchronous Agents are included in Visual C++ 10 CRT and are built on top of ConcRT, any workloads you build on them will scale to 256 cores--

just like that! (Well, you will still need to write the code, but the scaling is free.)

For managed code applications, the story is less bright. Management of the managed stack thread sits on top of the .NET ThreadPool (System.Threading.ThreadPool) by default. This does not use the new processor group APIs in Windows 7, and therefore doesn’t automatically benefit from the Windows 7 ability to scale. The maximum number of processes that the threadpool can utilize is 64. But not everything is lost, it is possible to write a custom TaskScheduler that targeted more than 64 procs, and use the rest of the Task Parallel Library with that special scheduler. This would be a cool CodePlex project, right?

Overall Visual Studio 2010 includes tons of new technologies and improvements--all of it icing on the Windows 7 cake!