You can configure your app to execute under the .NET 4 runtime by adding these lines to the executable’s configuration file, under the root  configuration element:

Because the security . . . → Read More: Running .NET 2 Runtime applications under the .NET 4 Runtime]]> In some situations, you might want to run a .NET 2 Runtime application (.NET 2, 3.0, 3.5 SP1 etc) under the .NET 4 Runtime – without recompiling.

You can configure your app to execute under the .NET 4 runtime by adding these lines to the executable’s configuration file, under the root  configuration element:

<startup>
  <supportedRuntime version="v4.0" sku=".NETFramework,Version=v4.0"/>
</startup>

Because the security framework has changed in the .NET 4 runtime, you will likely encounter some exceptions containing a message similar to:

System.NotSupportedException: This method explicitly uses CAS policy, which has been obsoleted by the .NET Framework. In order to enable CAS policy for compatibility reasons, please use the NetFx40_LegacySecurityPolicy configuration switch. Please see http://go.microsoft.com/fwlink/?LinkID=155570 for more information.

To avoid this, you  have to enable the legacy support by adding a runtime element :

<startup>
  <supportedRuntime version="v4.0" sku=".NETFramework,Version=v4.0"/>
</startup>
<runtime>
  <NetFx40_LegacySecurityPolicy enabled="true"/>
</runtime>

If your application also uses mixed assemblies that contain both managed and native code (such as System.Data.SQLite.dll), you’ll see an error message like this:

Mixed mode assembly is built against version ‘v2.0.50727′ of the runtime and cannot be loaded in the 4.0 runtime without additional configuration information.

You will need to enable the legacy activation to allow these to be loaded also:

<startup useLegacyV2RuntimeActivationPolicy="true">
  <supportedRuntime version="v4.0" sku=".NETFramework,Version=v4.0"/>
</startup>
<runtime>
  <NetFx40_LegacySecurityPolicy enabled="true"/>
</runtime>

References

<startup> Element @ MSDN

<supportedRuntime> Element @ MSDN

<NetFx40_LegacySecurityPolicy> Element @ MSDN

<startup useLegacyV2RuntimeActivationPolicy="true">
        <supportedRuntime version="v4.0" sku=".NETFramework,Version=v4.0"/>
    </startup>
    <runtime>
        <NetFx40_LegacySecurityPolicy enabled="true"/>
    </runtime>

Typically this involves hooking up some event handlers then waiting for some asynchronous code to fire the event before proceeding with the unit test and assertions.

The ManualResetEvent class (MSDN) seems a good choice for this, and this post has a small example of using it . . . → Read More: Unit testing multi-threaded, asynchronous code and/or events]]> I’ve been writing some unit tests recently that test some multi-threaded functionality.

Typically this involves hooking up some event handlers then waiting for some asynchronous code to fire the event before proceeding with the unit test and assertions.

The ManualResetEvent class (MSDN) seems a good choice for this, and this post has a small example of using it in a unit test:

[Test()]
public void AfterRunAsync()
{
    ManualResetEvent manualEvent = new ManualResetEvent(false);

    TestTestCase tc = new TestTestCase(1, "", 0, 0);
    bool eventFired = false;
    tc.RunCompleted +=
        delegate(object sender, AsyncCompletedEventArgs e) {
            Assert.IsInstanceOfType(typeof (TestTestCase), sender, "sender is TestCase");
            bool passed = tc.Passed;
            string output = tc.Output;
            eventFired = true;
            manualEvent.Set();
        };
    tc.RunAsync();
    manualEvent.WaitOne(500, false);
    Assert.IsTrue(eventFired, "RunCompleted fired");
}

The keys are:

uTorrent is one of the most popular BitTorrent clients out there. In my opinion it’s the best.

You can set up IP filtering in uTorrent to block bad seeds and peers from a list maintained by the community.

The assembly is signed and what is being done to the assembly is breaking the strong name. This is quite comforting to know; the strong name wouldn’t be . . . → Read More: Could not load file or assembly ‘x’ or one of its dependencies. Strong name validation failed.]]> In some of the work I’m doing right now, I’m manipulating an assembly after compile time – having it disassembled into IL, tweaked, then re-compiled back into an assembly.

The assembly is signed and what is being done to the assembly is breaking the strong name. This is quite comforting to know; the strong name wouldn’t be so strong if it was that easy to hack an assembly with a strong name.

When trying to load the hacked assembly, I am getting an exception (FileLoadException in this case but I’m guessing this may differ depending on your assembly load method) with the message “Could not load file or assembly ‘MyAssemblyName’ or one of its dependencies. Strong name validation failed.”.

The first interesting thing here is that the assembly named in the error message isn’t the hacked assembly; the hacked assembly is one of MyAssemblyName’s dependencies and is what’s triggering the error.

Make sure that you check the dependencies of the assembly named in the exception message when troubleshooting. The problem may be with one of the dependencies.

In my case the exception isn’t a surprise because of what’s being done to the assembly. But until I resolve that, how can I get around this for now?

You can exclude an assembly from strong name validation for development purposes using the Microsoft (R) .NET Framework Strong Name Utility tool aka sn.exe:

"%ProgramFiles%\Microsoft SDKs\Windows\v6.0A\bin\sn.exe" -Vr "C:\Path\To\Assembly.dll"

Make sure you change the sn.exe path depending on which version of the .NET Framework SDK you have installed. If you’re having trouble, get into the  ”%ProgramFiles%\Microsoft SDKs\Windows” dir and search for sn.exe, and use the newest one you can find.

You might find it handy to add this as a Post-build event command-line for your project from within Visual Studio in Project Properties > Build Events:

"%ProgramFiles%\Microsoft SDKs\Windows\v6.0A\bin\sn.exe" -Vr "$(TargetPath)"

So how do you switch the strong name validation back on for your assembly?

Use the -Vu switch:

"%ProgramFiles%\Microsoft SDKs\Windows\v6.0A\bin\sn.exe" -Vu "C:\Path\To\Assembly.dll"

A gotcha you must watch out for is caused by this piece of configuration from the NLog site: <configuration> <configSections> <section name="nlog" type="NLog.Config.ConfigSectionHandler, NLog"/> </configSections> <nlog> </nlog> </configuration>

Because you are not using the strong name for the Assembly-qualified name of . . . → Read More: Configuring NLog for your application when NLog is in the Global Assembly Cache (GAC)]]> If you have several applications that are using NLog, it can be a good idea to install NLog into the GAC and reference that.

A gotcha you must watch out for is caused by this piece of configuration from the NLog site:

<configuration>
<configSections>
<section name="nlog" type="NLog.Config.ConfigSectionHandler, NLog"/>
</configSections>
<nlog>
</nlog>
</configuration>

Because you are not using the strong name for the Assembly-qualified name of ConfigSectionHandler, it’s impossible to do a GAC lookup, therefore NLog won’t be found and you’ll get an application error (even if NLog is actually in the GAC).

This means your application will throw an exception when the configuration is loaded; there will be no NLog.dll in your application folder and it can’t check the GAC as it doesn’t have the strong name of the assembly you want.

You can fix this by including the strong name:

<section name="nlog" type="NLog.Config.ConfigSectionHandler, NLog, Version=1.0.0.505, Culture=neutral, PublicKeyToken=5120e14c03d0593c" />


Usually you specify the maximum height and width of the thumbnail, e.g. 150 x 200.

However, if your original image’s aspect ratio is different to the maximum thumbnail dimensions, you will end up with dead space vertically or horizontally (shown in green in the illustration). This . . . → Read More: Creating thumbnail images with automatic cropping and maintaining aspect ratio]]> Most thumbnail-generation solutions will shrink the original down while maintaining aspect ratio.

Usually you specify the maximum height and width of the thumbnail, e.g. 150 x 200.

However, if your original image’s aspect ratio is different to the maximum thumbnail dimensions, you will end up with dead space vertically or horizontally (shown in green in the illustration). This can be quite an eyesore when displaying thumbnails in a grid.

I’ve got an algorithm that will automatically crop the image either horizontally or vertically to then match the thumbnail aspect ratio, so you end up with the thumbnails all being the same size even though they may be coming from originals of wildly different aspect ratio.

In the illustration, you can see that the image is scaled down and fills all the available thumbnail space, showing the parts in grey from the original that were cropped out.

The algorithm does the cropping before the resizing. It takes the width and height of the original image, and the width and height of the desired thumbnail. It will return the width and height that the original must be cropped to, to match the aspect ratio of the thumbnail.

Of course this isn’t quite enough; where do we do the cropping? All from the top of the image, or the bottom, or a mix? Ideally we want the user to choose which parts of the original they want cropped out. But we can do some very simple automatic cropping by cropping equal amounts from each side that needs to be cropped.

Now we know exactly which part of the original we can take and then resize down to a thumbnail.

http://hammett.castleproject.org/?p=336

While I’m currently focusing my efforts on ASP.NET MVC and moved away . . . → Read More: Castle Project update]]> Hamilton has applied a bunch of patches and fixes to the Castle Project for the Components and Windsor sub-projects. Including 2 minor and insignificant patches that I submitted +D. It’s good to feel part of a major Open Source project, even for a small contribution.

http://hammett.castleproject.org/?p=336

While I’m currently focusing my efforts on ASP.NET MVC and moved away from MonoRail, I use Castle extensively in my .NET projects, particularly ActiveRecord and Windsor/MicroKernel.

VS is a resource hog, so it’s not long before my laptop is struggling having a reasonable size solution in addition to the usual open applications.

There are a few things you can do to try and squeeze a bit more performance out of Visual . . . → Read More: Visual Studio Performance]]> I’m running Visual Studio 2008 on a old, single-core laptop with 1GB RAM.

VS is a resource hog, so it’s not long before my laptop is struggling having a reasonable size solution in addition to the usual open applications.

There are a few things you can do to try and squeeze a bit more performance out of Visual Studio I’ve found from various blogs and looking around myself. The first two are the most useful. Don’t muck with the other options unless you know what they’re going to do.

Recycle Memory
Perhaps the biggest trick is the simplest: if you minimize Visual Studio, it will force a garbage collection, and packing of the memory usage. This can make VS drop from a few hundreds MB of memory usage to 20 or less. Of course things will slow down when you go back into VS as things get reloaded into memory from virtual memory and disk.

Patch to SP1
Make sure Visual Studio 2005 SP1 is installed, for the bug and performance fixes since release.

Turn off animation
Tools > Options > Environment > General and uncheck Animate environment tools.

Disable the Navigation Bar
Go to Tools > Options > Text Editor > C# and uncheck Navigation bar under Display.

Turn off Track Changes
Tools > Options > Text Editor and uncheck Track changes.

Turn off Track Active item
Tools > Options > Projects and Solutions: uncheck Track Active Item in Solution Explorer.

Turn off AutoToolboxPopulate.
Tools > Options > Windows Forms Designer: set AutoToolboxPopulate to False.

Startup with an empty environment
Tools > Options > At startup: Show empty content

Windows Forms refactoring
Tools > Windows Forms Designer > General: Set “EnableRefactoringOnRename” to false in the Refactoring section

I have a nice continuous integration server going smoothly, but my most recent check-in threw up a strange error when running the Unit Tests:

System.BadImageFormatException : Could not load file or assembly ‘Overclockers.Web, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null’ or one of its dependencies. The module was expected to contain . . . → Read More: System.BadImageFormatException]]> I’m currently build the new Overclockers New Zealand website in .NET

I have a nice continuous integration server going smoothly, but my most recent check-in threw up a strange error when running the Unit Tests:

System.BadImageFormatException : Could not load file or assembly ‘Overclockers.Web, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null’ or one of its dependencies. The module was expected to contain an assembly manifest.

The server is actually a VMWare instance, but is running a 32-bit version of Windows Server 2003 EE.

After some troubleshooting and searching, I decided to first try to explicitly build for x86 first anyway. This is done in the Project Properties dialog in Visual Studio.

When changing this, make sure you make the change for both Debug and Release configurations.

This seemed to fix the problem.

I suspect this could be because I have some Shell32 interop in one of my shared libraries, and this may have caused problems, and/or the fact I’m running the instance on VMWare.