Rediger

Del via


Managed Types (C++/CLI)

Visual C++ allows access to .NET features through managed types, which provide support for features of the common language runtime and are subject to the advantages and restrictions of the runtime.

Managed Types and the main Function

When you write an application using /clr, the arguments of the main() function can't be of a managed type.

An example of a proper signature is:

// managed_types_and_main.cpp
// compile with: /clr
int main(int, char*[], char*[]) {}

.NET Framework Equivalents to C++ Native Types

The following table shows the keywords for built-in Visual C++ types, which are aliases of predefined types in the System namespace.

Visual C++ type .NET Framework type
void System.Void
bool System.Boolean
signed char System.SByte
unsigned char System.Byte
wchar_t System.Char
short and signed short System.Int16
unsigned short System.UInt16
int, signed int, long, and signed long System.Int32
unsigned int and unsigned long System.UInt32
__int64 and signed __int64 System.Int64
unsigned __int64 System.UInt64
float System.Single
double and long double System.Double

For more information about the compiler option to default to signed char or unsigned char, see /J (Default char type is unsigned).

Version Issues for Value Types Nested in Native Types

Consider a signed (strong name) assembly component used to build a client assembly. The component contains a value type that is used in the client as the type for a member of a native union, a class, or an array. If a future version of the component changes the size or layout of the value type, the client must be recompiled.

Create a keyfile with sn.exe (sn -k mykey.snk).

Example

The following sample is the component.

// nested_value_types.cpp
// compile with: /clr /LD
using namespace System::Reflection;
[assembly:AssemblyVersion("1.0.0.*"),
assembly:AssemblyKeyFile("mykey.snk")];

public value struct S {
   int i;
   void Test() {
      System::Console::WriteLine("S.i = {0}", i);
   }
};

This sample is the client:

// nested_value_types_2.cpp
// compile with: /clr
#using <nested_value_types.dll>

struct S2 {
   S MyS1, MyS2;
};

int main() {
   S2 MyS2a, MyS2b;
   MyS2a.MyS1.i = 5;
   MyS2a.MyS2.i = 6;
   MyS2b.MyS1.i = 10;
   MyS2b.MyS2.i = 11;

   MyS2a.MyS1.Test();
   MyS2a.MyS2.Test();
   MyS2b.MyS1.Test();
   MyS2b.MyS2.Test();
}

The example produces this output:

S.i = 5
S.i = 6
S.i = 10
S.i = 11

Comments

However, if you add another member to struct S in nested_value_types.cpp (for example, double d;) and recompile the component without also recompiling the client, the result is an unhandled exception (of type System.IO.FileLoadException).

How to test for equality

In the following sample, a test for equality that uses Managed Extensions for C++ is based on what the handles refer to.

Example

// mcppv2_equality_test.cpp
// compile with: /clr /LD
using namespace System;

bool Test1() {
   String ^ str1 = "test";
   String ^ str2 = "test";
   return (str1 == str2);
}

The IL for this program shows that the return value is implemented by using a call to op_Equality.

IL_0012:  call       bool [mscorlib]System.String::op_Equality(string, string)

How to diagnose and fix assembly compatibility problems

When the version of an assembly referenced at compile time doesn't match the version of the assembly referenced at runtime, various problems may occur.

When an assembly is compiled, other assemblies may be referenced with the #using syntax. During the compilation, these assemblies are accessed by the compiler. Information from these assemblies is used to make optimization decisions.

However, if the referenced assembly is changed and recompiled, also recompile the referencing assembly that's dependent on it. Otherwise, the assemblies might become incompatible. Optimization decisions that were valid at first might not be correct for the new assembly version. Various runtime errors might occur because of these incompatibilities. There's no specific exception produced in such cases. The way the failure is reported at runtime depends on the nature of the code change that caused the problem.

These errors shouldn't be a problem in your final production code as long as the entire application is rebuilt for the released version of your product. Assemblies that are released to the public should be marked with an official version number, which will ensure that these problems are avoided. For more information, see Assembly Versioning.

To diagnose and fix an incompatibility error

You may encounter runtime exceptions or other error conditions in code that references another assembly. If you can't identify another cause, the problem may be an out of date assembly.

  1. First, isolate and reproduce the exception or other error condition. A problem that occurs due to an outdated exception should be reproducible.

  2. Check the timestamp of any assemblies referenced in your application.

  3. If the timestamps of any referenced assemblies are later than the timestamp of your application's last compilation, then your application is out of date. If it's out of date, recompile your application with the most recent assemblies, and edit your code if necessary.

  4. Rerun the application, perform the steps that reproduce the problem, and verify that the exception doesn't occur.

Example

The following program illustrates the problem: it first reduces the accessibility of a method, and then tries to access that method in another assembly without recompiling. Compile changeaccess.cpp first. It's the referenced assembly that will change. Then compile referencing.cpp. It should successfully compile. Next, reduce the accessibility of the called method. Recompile changeaccess.cpp with the compiler option /DCHANGE_ACCESS. It makes the access_me method protected, rather than public, so it can't be called from outside Test or its derivatives. Without recompiling referencing.exe, rerun the application. A MethodAccessException occurs.

// changeaccess.cpp
// compile with: /clr:safe /LD
// After the initial compilation, add /DCHANGE_ACCESS and rerun
// referencing.exe to introduce an error at runtime. To correct
// the problem, recompile referencing.exe

public ref class Test {
#if defined(CHANGE_ACCESS)
protected:
#else
public:
#endif

  int access_me() {
    return 0;
  }

};

Here's the source for the referencing assembly:

// referencing.cpp
// compile with: /clr:safe
#using <changeaccess.dll>

// Force the function to be inline, to override the compiler's own
// algorithm.
__forceinline
int CallMethod(Test^ t) {
  // The call is allowed only if access_me is declared public
  return t->access_me();
}

int main() {
  Test^ t = gcnew Test();
  try
  {
    CallMethod(t);
    System::Console::WriteLine("No exception.");
  }
  catch (System::Exception ^ e)
  {
    System::Console::WriteLine("Exception!");
  }
  return 0;
}

See also

.NET programming with C++/CLI (Visual C++)
Interoperability with other .NET languages (C++/CLI)
Managed types (C++/CLI)
#using directive