CORBA technology enables an object in one process to call the methods
of another object running in another process. The basic parts of programs
that use CORBA technology are shown in this
The roles that each class play in a CORBA program are summarized here:
Client class. Calls a method that will be executed in the server
process. Note that a complete client-side application might consist of
more classes than the CORBA client class.
Stub classes. Converts the client class' method invocation
into a format that can be delivered over a network connection. Used on
the client side.
ORB classes. Manages the transmission and delivery of method
invocations between processes.
Implementation class. Contains the server's business logic. An instance
of this class, known in CORBA terms as the servant, is registered
with the ORB. Methods of this instance can be executed by clients running
in other processes.
Server class. Instantiates the implementation class (creates the
servant). The server class also takes a reference to the servant
instance and makes it publicly available by registering the reference with
a naming service, or by writing the reference to standard output, by
writing the reference to a specified file. This reference is known in CORBA
terms as interoperable object reference (IOR).
Skeleton classes. Converts a method invocation from the wire
format generated by the stub classes and transmitted by the ORB into a
format that can be read by the servant instance. Used on the server
side.
CORBA is a complex technology and a comprehensive account of it is
outside the scope of this document. If you are developing CORBA programs
and you are not familiar with the technology, you can look at any of
the books that have been written on CORBA.