Working with C#, C++/CLI and Managed C++

Recently I spent nearly 2 months working on a windows project mixing different languages such as C#, C++/CLI, Managed C++. Here is what I feel:

• Writing in C# is more fun and efficient (developing time, not execution time). You can use automatic property, anonymous function with lambda expression and LINQ. The code is more concise and expressive.

• C++/CLI is really great to mix native type and managed type together. But sometimes it takes too much brain energy (or think about) to jump between the two worlds. Also, I don’t really like way that the generated UI code is in the same file as the header file, it makes the header file a really mess! Anyway, here are some useful resources/tips I can share:

  • C++/CLI Cheat Sheet
  • C++/CLI in Action
  • Standard ECMA-372: C++/CLI Language Specification
  • For C# developers, don’t forget that you can and you should use C++ Interop instead P/Invoke. It provides better performance.

• Managed C++: I’m really happy Microsoft has deprecated them. It’s the predecessor of C++/CLI and has all those weird syntax: __gc, __box and confusing symbols: * for both native and managed type. Coding between it and C++/CLI is really a brain f**k. At a point, I have to convert one project from Managed C++ to C++/CLI to really gets speed up. The conversion process is not as slow as I thought, largely due to the power of VIM (there is a nice Visual Studio extension ViEmu) and some regex magic.

For new applications, I’d suggest the following stack:

• (If necessary) use C++/CLI to interop with native C++ libraries and provide core functionality for the program.

• Use C# to code the UI part.

• Code the application’s core part first, make it work without any GUI: it seperates the business logic and makes the code more testable.

• Use data binding, don’t code too much event handler in your GUI.