接口与实现分离
为什么这样设计？
主要原因是保持接口的稳定，而且封装性更好。类的实现细节跟其他类的联系都隐藏起来了。
具体实现
Database.h
Create的实现其实是调用Impl的实现。但在头文件中只需要CDatabaseImpl的声明。
代码
1 #ifndef _DATABASE_H_
2 #define _DATABASE_H_
3 class CDatabaseImpl;
4 class CDatabase
5 {
6 public :
7 CDatabase();
8 ~ CDatabase();
9 int Create( const char * pName);
10 private :
11 CDatabaseImpl * m_pImpl;
12 };
13
14 #endif
Database.cpp
代码
#include " Database.h "
#include " DatabaseImpl.h "
CDatabase::CDatabase()
{
m_pImpl = new CDatabaseImpl;
}
CDatabase:: ~ CDatabase()
{
if (m_pImpl)
delete m_pImpl;
m_pImpl = NULL;
}
int CDatabase::Create( const char * pName)
{
return m_pImpl -> Create(pName);
}
DatabaseImpl.h
代码
#ifndef DATABASE_IMPL_H
#define DATABASE_IMPL_H
#include < stdio.h >
class CDbMainPart
{
public :
int CreateMain()
{
printf( " Main Created\n " );
}
};
class CDbMinorPart
{
public :
int CreateMinor()
{
printf( " Minor Created\n " );
}
};
class CDatabaseImpl
{
public :
CDatabaseImpl(){}
~ CDatabaseImpl(){}
int Create( const char * pName);
private :
CDbMinorPart m_MinorPart;
CDbMainPart m_MainPart;
};
#endif
DatabaseImpl.cpp
代码
}#include < stdio.h >
#include " DatabaseImpl.h "
int CDatabaseImpl::Create( const char * pName)
{
printf( " [%s] Created\n " ,pName);
m_MainPart.CreateMain();
m_MinorPart.CreateMinor();
return 0 ;
main.cpp
#include " Database.h "
int main()
{
CDatabase db;
db.Create( " Longshine " );
return 0 ;
}