I need the help of the C++ gurus out there.  I have the following code, it's
working great, but I'm concern about the lack memory management I'm doing.  I
just want to confirm if I'm doing something wrong or my code is fine.  I
don't want to put this code in production and find out in some days that my
servers are going to crash.

I have the need to integrate an old system, written in C with .NET.  What
I'm doing is using an intermidiate C++.NET class as a bridge to get benefit
of the IJW technology, here is a sample code:

#include "stdafx.h"
#include "MyCPPClass.Interop.h"

#pragma managed
#using <mscorlib.dll>
#using "MyCSharpAssembly.dll"
using namespace System;
using namespace System::Runtime::InteropServices;
using namespace MyCSharpAssembly;

void ExecuteManageCode(char *msg, char **errMsg)
{
 try{
   MyCSharpAssembly::CSharpClass *cSharpClass = new
MyCSharpAssembly::CSharpClass();
   cSharpClass->set_Msg(msg);
   cSharpClass->Execute();
 }
 catch(Exception *e){
   *errMsg =
(char*)Marshal::StringToHGlobalAnsi(e->get_Message()).ToPointer();
 }
}

#pragma unmanaged
extern "C" {
 __declspec(dllexport) void __stdcall ExecuteUnManageCode(char *msg, char
**errMsg)
 {
   ExecuteManageCode(msg, errMsg);
 }
}

I modify the original code for simplicity, but this is pretty much what I'm
doing.  My concern is if I should worry about destroying the objects created
in the MANAGED section:

   MyCSharpAssembly::CSharpClass *cSharpClass = new
MyCSharpAssembly::CSharpClass();

or the GC should take care of that, and the other question is that in the
UNMANAGED section, should I do anything important here, or it's pretty much
fine?

NOTE:  I noticed that after excuting my code, if the C application(client)
is running I can't override or replace the C++.NET assembly, seems to be
"lock" by my C application, no matter if the execution to the method finished
already,  is this OK?

Thanks for your help