The process is explained here:
http://msdn.microsoft.com/en-US/library/d1ae6tz5%28v=VS.80%29.aspx
What I don’t get from that article is that the pinned wchar_t* is passed to various C string functions that rely on the trailing null character. Is it a rule that .NET strings have a trailing null character? The System.String docs say that:
In the .NET Framework, a null character can be embedded in a string. When a string includes one or more null characters, they are included in the length of the total string.
From what I understand, the internal character buffer in .NET CLR strings is null-terminated, although naturally that null character doesn’t make it into the character count, so any .NET code would ignore it. The only reason that null is there is for easy interop with Windows API or other plain C code that expects strings to be null-terminated. Instead of appending a null character any time a .NET string has to be passed to a C API (and possibly having to reallocate and copy the entire string), the null character is just there from the first place – a useful optimization in the real-world, since .NET still has to do a lot of interop behind the scenes, even if you don’t use it explicitly.
If you do happen to have some null characters in the middle of your string, well, any C API that gets your string would probably stop there and never reach the end of the string. I guess you can even try it in C++/CLI yourself and see what happens. 🙂