The following sequence summarizes the typical steps in the management of a named pipe:

• The server process creates the pipe by calling DosCreateNPipe. DosCreateNPipe returns a handle for the server end of the pipe. (Note that the server uses the same handle to both read from and write to the pipe.) The pipe is now in the disconnected state and cannot be opened by a client process. The server process calls DosConnectNPipe to put the pipe into a listening state.

• The pipe can now be opened by a client process.

• A client process supplies the name of the pipe in a call to DosOpen and receives a pipe handle in return. (The client uses the same handle to both read from and write to the pipe.) The pipe is now in the connected state and can be read from or written to by the client.

• The server and client processes communicate by calling DosRead and DosWrite. DosResetBuffer can be used to synchronize read and write dialogs. A server process that supports a large number of clients for a local named pipe can use DosSetNPipeSem and DosQueryNPipeSemState to coordinate access to the pipe. Server and client processes can also use DosTransactNPipe and DosCallNPipe to facilitate their communication.

• After completing its transactions, the client process calls DosClose to close its end of the pipe. The pipe is now in the closing state and cannot be accessed by another client.

• The server process calls DosDisConnectNPipe to acknowledge that the client has closed its end of the pipe. The pipe is now in the disconnected state again.

• To enable another client process to open the pipe, the server must call DosConnectNPipe again. This puts the pipe back into the listening state. To end its access to the pipe, the server calls DosClose. When all of the handles for both ends of the pipe have been closed, the pipe is deallocated by the system.