c++/cli 托管 非托管类型 转换 marshal_as（vs2008） marshal_to（vs2005）

#include <msclr/marshal.h>

using namespace msclr::interop;

msclr::interop::marshal_context^ context;

const wchar_t* __MarshaledReturnVal = context->marshal_as<const wchar_t*>(__ReturnVal);

I spent quite some time now doing research on C++ to .NET bridging
using C++/CLI and found your approach and code to be the cleanest from
all the examples I looked at. However, getting it to work required some
changes which made me wonder if you ever tried compiling above code
samples. Or maybe the way I did it was different than intended?

For anyone interested in my working solution, here my approach:

1. I used Visual Studio 2008 and created 3 projects:

- A C# class library called ManagedClasses which contains the Calculator class.

- A C++ class library with CLR support (/clr) called ILBridge which
contains the ILBridge_CppCliWrapper_Calculator and
NativeExport_CppCliWrapper_Calculator classes.

- A C++ console application without CLR support called NativeApp which uses the ILBridge lib.

Due to changes in the way the marshaling between objects is done in
Visual Studio 2008 I added an include for <msclr/marshal.h> and
changed the code in the ILBridge class from this line:

wchar_t* __MarshaledReturnVal = marshal_to<wchar_t*>(__ReturnVal);

to this code snippet:

msclr::interop::marshal_context^ context;

const wchar_t* __MarshaledReturnVal = context->marshal_as<const wchar_t*>(__ReturnVal);

delete context;

Also, I had to split up the ILBridge class into header and source
file, otherwise the NativeExport class would not compile, because the
compiler kept complaining that you cannot use marshal_as in a class that
is translated into native code.

Then I added

#using "../ManagedClasses/bin/Debug/ManagedClasses.dll"

to the ILBridge class header so that it finds the C# assembly.

Finally I took out the #ifdef THISDLL_EXPORTS from the NativeExport
class header, since unless you define this symbol it won't set the
correct dll export.