The class sp_name has a member LEX_STRING m_qname. This member is needed only for error reporting purposes. it contains redundant information and is always synchronized with the other two members LEX_STRING members
m_db and m_name. In normal cases (when no errors happen) the current code assumes redundant work to keep m_qname valid whenever m_db and m_name are initially set or are later modified.

Instead of maintaining the redundant member m_qname we'll introduce two new classes:

1. Database_qualified_name

class Database_qualified_name

{

public:

  LEX_STRING m_db;

  LEX_STRING m_name;

  ...

}

Under terms of this task we'll derive sp_name and sp_head from this new class Database_qualified_name.

Later, under terms of separate tasks, we'll derive more classes from Database_qualified_name. For example, Item_ident is a good candidate for.

2. ErrConvDQName - a new ErrConv family class (in addition to those defined in sql_error.h}
This class will be used whenever we need to send a database qualified object name to an error message.

class ErrConvDQName: public ErrConv

{

  const Database_qualified_name *m_name;

public:

  ErrConvDQName(const Database_qualified_name *name)

   :m_name(name)

  { }

  const char *ptr() const

  {

    // code to make a string in format 'dbname.name' in {{ErrConv::err_buffer}}

  }

};

Open question:
Perhaps the class name should be Database_qualified_object_name. Because later we'll need a derived class Database_qualfieid_column_name, which will add a new LEX_STRING member to store column names.