Messages serve many purposes: to provide information, to ask a question and request a reply, and to solicit some action on the recipient's part.

Editor's note: This article is an excerpt from Chapter 16, "Advanced Message Handling," of Control Language Programming for IBM i.

Understanding how messages work is an important part of writing good CL programs. In fact, without at least a basic understanding of messages, your CL programs will be error-prone and, in many cases, unreliable.

IBM i is driven by messages. Messages are used for a multitude of different purposes, including initiating jobs, executing commands, communicating between programs, signaling error conditions, letting users communicate with each other, and letting jobs communicate with users.

What Is a Message?

In everyday life, the concept of a message is quite simple. Webster's dictionary defines a message as "a communication, verbal or written, sent by one person to another." A message, by definition, is sent and is not of any true significance until it is received. Once a message has been received, it can be acted upon. The one who receives the message decides what to do with the information contained in the message.

Messages serve many purposes: to provide information, to ask a question and request a reply, and to solicit some action on the recipient's part. Regardless of the message type, the recipient determines what will be done after the message has been received. The receiver can ignore the message, act on the message, or, if requested, choose to send a reply.

IBM i messages are no different: A message is sent to convey information, and the recipient of the message determines what will be done with that information.

Message Queues

Messages can be sent from only two sources: system users and running programs. Messages are not sent directly to other users or programs but rather to a message queue associated with a user, a program, or a workstation device.

A message queue (object type *MSGQ) is a holding area or "in basket" for messages. A message queue usually can store hundreds or even thousands of messages. Once a message has arrived on a message queue, it is up to the recipient to view (or receive) the message, take any action, and remove the message when it is no longer needed.

When a user profile object is created, a user message queue is created for that user. The system uses this message queue whenever a message is sent specifically to that user. Usually, the name of the message queue object is the same as the name of the associated user profile.

Whenever a workstation device is created, a workstation message queue is created for that device. Some types of messages are routinely sent to a workstation message queue rather than to a user message queue. It is the responsibility of the workstation user to view any messages sent to the workstation message queue and to take any action necessary. The workstation message queue is usually named the same as the associated workstation device.

In addition, a few special message queues exist on every system. The system operator message queue (QSYSOPR) lets the system operator receive messages sent by the system, by programs, or by other users and take appropriate action. The system sends system-related messages to the system history log message queue (QHST), which you then can view using the DSPLOG (Display Log) command. You can create other message queues as needed using the CRTMSGQ (Create Message Queue) command.

All the message queues we've discussed so far are permanent objects on the system. You can delete them using the DLTMSGQ (Delete Message Queue) command, but they will be permanent until explicitly deleted. The IBM i operating system also supports two types of temporary message queues: a call message queue and an external message queue.

Immediately before a program begins executing, the system creates a message queue for that program. The call stack entry message queue (sometimes called the call message queue or the program message queue) is the program's own "in basket" for messages. Every called program has an associated call stack entry message queue. In an ILE CL program, each procedure in the program also has its own call message queue. Once a message arrives on the call message queue, it is up to the program or procedure to determine what to do with it. Certain messages can be ignored; others must be dealt with, depending on the message type.

Whenever a job starts, an external message queue is created for that job. The external message queue is used to communicate between the executing job and the job's external requester. Many different message types can be sent to the external message queue as a job runs. Some require attention; others do not.

The structure of all message queues is the same, whether they are permanent or temporary. There is an area for the message and an area for a system-generated message key. The system uses the message key to uniquely identify each message on the queue. The message storage area is used to hold the message type, severity, text, and other data associated with the message.

Types of Messages

Several message types can be sent and received. Every message in a message queue has an associated type, such as *DIAG, *COMP, *INFO, *INQ, and *ESCAPE. Each message type has a unique purpose; the message type sent will depend on the purpose of the message as determined by the sender of the message. The following table lists the types of messages, a description of each type, and its purpose.

You must be aware of certain rules about sending and receiving different types of messages. A workstation user can send only *INFO, *INQ, and *RPY message types. A CL program can send any of the message types. The following table summarizes the types of message queues and which message types they can accept.