Sei sulla pagina 1di 4

ACTIVATION GROUPS

There are three types of activation groups: 1. The default activation group (*DFTACTGRP) 2. Persistent or named activation groups (also called static activation groups) 3. Non-persistent or *NEW activation groups (also called dynamic activation groups) Activation groups are boundaries between running programs/applications within one job. One or more programs can run within an activation group, and there can be one or more activation groups for any job. Activation groups are assigned to each compiled program or service program at compile time, using the ACTGRP parameter of the CRTPGM, CRTSRVPGM, or CRTBNDxxx command. With the exception of the *DFTACTGRP, activation groups are automatically created when a program is called. An activation group name is associated with all program objects. When a program is called, if its associated activation is not active, the system creates the activation group and then loads the program. If the activation group is already activated, the program is loaded immediately. This provides much faster program loading.

Default Activation Group


When a job is started on the iSeries or AS/400, *DFTACTGRP is created, and it cannot be terminated until the job ends. *DFTACTGRP is where all original OS/400 program objects run. For example, RPG III and CL programs run in *DFTACTGRP along with QCMD, QCMDEXC, PDM, and most other system programs. However, the ILE debugger runs in a named activation group. In addition, limited-function RPG IV programs can run in *DFTACTGRP. Limited function is defined as programs that don't contain any procedures, don't call any procedures, and use no contemporary built-in functions (BIFs) as well as any features that require procedures not supported by *DFTACTGRP. Effectively, an RPG III program that is converted to RPG IV with the CVTRPGSRC command is considered a limited-function RPG IV program. These programs can run in *DFTACTGRP or any other activation group.

Named Activation Groups


A named activation group is an activation group that is created by the system when a program associated with it is run. Named activation groups are persistent--that is, they do not close down when the last program in the invocation stack has ended. This does two things: (1) It allows the program to be subsequently loaded more quickly; the activation group does not have to be reactivated, and (2) It allows other programs that are compiled with the same activation group name to be loaded into the activation group.

*NEW Activation Groups


*NEW activation groups are created when a program compiled with ACTGRP(*NEW) is called. In this situation, the system generates a temporary activation group name, activates the activation group, and loads the program into it. Other programs called by the program can be loaded in this activation group by specifying ACTGRP(*CALLER) when they are compiled. Each time the program with ACTGRP(*NEW) is called, a new activation group is created and the program is loaded into that activation group. This implies that recursive calls are supported under ILE. This is a correct assumption. *NEW activation groups are not persistent; they automatically close down when the program that started them ends.

Assigning an Activation Group


Activation groups are assigned to programs and service program when the program or service program is created. The ACTGRP parameter of the CRTPGM and CRTBNDxxx commands identifies the name of the activation group in which the program runs. The values for the ACTGRP parameter are as follows: *NEW--This is the default for the CRTPGM command. It causes the system to create a unique, new activation group name whenever the program is called. In fact, if the program is called multiple times in the same job, a new activation group is created each time. If the program is called recursively, a new activation group is started for each invocation of the program. When the program ends, the new activation group is automatically destroyed. This occurs regardless of the status of the *INLR indicator in RPG IV programs. *CALLER--When a program with an activation group of *CALLER is called, it is loaded into the activation group of the program that called it. No new activation group is created. Whether that calling program's activation group is *NEW or a named activation group is irrelevant. *CALLER is useful for job streams in which you may have programs that are used in multiple applications and you want them to all run in the same activation group. Named Activation Group --This allows you to give an activation group a specific name. That name is used as the activation group in which the program is run. If the activation group is not already active, it is activated when the program is called. If the activation group is already active, the program is loaded into it. Use any name you like as the name of the activation group.

The CRTBNDRPG Command


The CRTBNDRPG command (PDM option 14) allows you to control the activation group for the program. Unlike the CRTPGM command, CRTBNDRPG allows you to specify whether the program runs in *DFTACTGRP or a real activation group. By real activation group, I mean a *NEW, *CALLER, or named activation group. CRTBNDRPG supports this capability with two parameters: DFTACTGRP and ACTGRP.

DFTACTGRP (*YES | *NO) Parameter


*YES--The program runs in *DFTACTGRP. The program is checked for any features that are not compatible with *DFTACTGRP. *YES is the default for the this parameter, so if this option is selected, the ACTGRP parameter is ignored.. *NO--The activation group name specified on the ACTGRP parameter is assigned to the program.

ACTGRP (QILE | activation-group-name | *NEW | *CALLER)


QILE is the default name for a named activation group on the CRTBNDxxx command. This name is simply a default and has no implicit value. For simplicity, you might consider using the QILE activation group name until you become familiar with activation groups. Any name may be specified for the activation group.

Starting and Ending Activation Groups


The activation group name associated with a program is automatically started when the program is called. There is no Start Activation Group command, although sometimes I wish there were. The methods you can use to end an activation group depend on which activation group you're using. There is no way to end *DFTACTGRP. It ends when the job ends. *NEW activation groups, however, take care of themselves. They are automatically destroyed when the program that started the *NEW activation group returns to its caller. This deactivation is performed regardless of the setting in RPG IV programs of the *INLR indicator. When a program running in a *NEW activation group is called a second time, a new copy of it is loaded into memory and run. Named activation groups are handled differently. Although RPG IV provides no integrated function for terminating a named activation group, there is an API (CEETREC) that allows RPG IV or any ILE language to terminate its activation group. One of the advantages of named activation groups is also one of the biggest disadvantages--they are persistent. That is, they do not go away until the job in which they are active has ended. Sometimes, such as during development, is it important to end the named activation group so that you can load a new/fresh copy of the program on the next call to the program. IBM provides several methods to end an activation group, including the following: Use Use Call End the Reclaim Activation Group (RCLACTGRP) command a language feature such as STOP RUN in COBOL an ILE API the job

The RCLACTGRP Command


The RCLACTGRP command can be used to terminate (close down) inactive activation groups. Inactive activation groups are named activation groups that are no longer in use. The activation group name being closed down is specified on the ACTGRP parameter of the RCLACTGRP command. Optionally, you can end all named activation groups by specifying ACTGRP(*ELIGIBLE), as shown below:
RCLACTGRP *ELIGIBLE

While "eligible" may not be a familiar word to the CL language, you will need to get used to it. The following is another example of the RCLACTGRP command:
RCLACTGRP ACTGRP(QILE)

This command will end the QILE activation group, provided all program invocations have returned. An example of where QILE would not end is if QCMD is called from a program that is running within QILE; then, the QILE activation group is not eligible for termination.

Using STOP RUN in COBOL


RPG programmers tend to not use other programming languages, so COBOL may not be installed on your system. COBOL, however, has a longstanding STOP RUN statement. This ends the process for the set of COBOL procedures. In ILE, STOP RUN also ends the named activation group and returns to the caller. Unfortunately, RPG IV does not have a similar function.

ILE APIs
Along with ILE, IBM provided several APIs that perform certain tasks. One API, CEETREC, causes the current activation group to be deactivated, and all programs running in that activation group are terminated. This is effectively the same behavior as the STOP RUN statement in COBOL. The CEETREC API is an ILE procedure, and, therefore, must be called with a bound call operation--either a CALLB or CALLP to a prototype that references CEETREC. RPG IV and CL support bound calls, so they can call CEETREC. In RPG IV, a bound call can be performed with either a prototype call (CALLP) or the CALLB operation code. A prototype call, as the term implies, requires a prototype. While a prototype may be useful for completeness, it isn't necessary to call CEETREC.
C Eval *INLR = *ON C CallB 'CEETREC'

Setting on LR does not impact the results. CEETREC can be called with *INLR on or off. Regardless of the setting, the program is still removed from memory.

Clear as Mud?
Activation groups are complex, and this article only scratches the surface. Use them progressively, adding more to your applications as you become accustomed to using them. And don't feel bad if you don't understand them right away. A group of us were in Rochester, Minnesota, several years ago for an introduction to ILE. We developed a pet name for this new technology; we called them aggravation groups.

Potrebbero piacerti anche