.NET GUI - C# vs C++/CLI

Question

I'm writing a small app that requires a few listboxes, buttons, textboxes. It'll be linked with Boost, MySQL, etc. C++ static libs. The project requires win32 functions. I figure Winforms will be fine (MFC and CodeJock require too much time).

So C++/CLI seems perfect for the job. Just use standard C++ along side the GUI. Then I run across threads suggesting you write your GUI in C# instead. Then use p/Invoke (slow) or a C++/CLI interface to your standard C++ DLL's.

Example: http://social.msdn.microsoft.com/Forums/en-US/clr/thread/6ae877ac-07b4-4d26-8582-de475ee9a5cb

Why? What advantage is there in using C# for your winforms GUI instead of C++/CLI (they look the same, the commands are the same). What disadvantage is there in using C++/CLI executable instead of standard C++ executable. I can understand if cross-platform compatibility was an issue, but then you could simply not use managed features (other than the GUI).

I don't understand why you would use C#, and then go so far to separate it with an "engine DLL". Unless of course the "engine DLL" was being used for other applications as well.

Thanks

Answer 1

I think most recommendations with regard to this question center around the fact that C# is just a better environment to create .NET applications with than C++/CLI. The syntax is cleaner, the tooling is better - both within Visual Studio and from 3rd parties. You will get more and better support from developers who will almost all be more familiar with C#.

C++/CLI applications are different enough from standard C++ with all those ^ and % characters that I at least feel it is NOT C++.

Most advice is also coming from the point of view that you want to create a .NET application and C++/CLI is used more as a glue layer. Whenever I have used C++/CLI, it was grudgingly and almost always because some third-party library had many complex C/C++ objects that it passed around. When using C# and P/Invoke, you often have to create classes to mirror the structs and classes that are in the C++ header files of the software you are interfacing with. Keeping those in sync is labor intensive and making mistakes is easy to do. Furthermore, figuring out how to marshal a struct with pointers to structs of arrays of struct will make your brain melt!

My general advice is to use C# (or VB.NET) to create as much code as feasible for your application. Use P/Invoke when your need to call the Win32 API and/or 3rd party SDKs is limited and the interfaces and parameters are simple. Use C++/CLI as a glue layer when this is not possible.

In a team environment, your fellow developers will thank you for limiting your usage of C++/CLI to only where it is absolutely, positively required. C++/CLI expertise is just not that common.