Data Management

A data item is a value of a variable or constant. Data items can be single items called scalars, or a group of items called a data aggregate. A data aggregate has two types: arrays and structures.

Data used in a PL/I program is either computational data or program-control data.

In the following example, VAR1 and VAR2 are coded arithmetic variables and the numbers, 2 and 3.14, are coded arithmetic constants:

Program-control data constants are determined by a compiler.

In the following example, LAB1 is a label constant of program-control data. The LAB1 value is the address of "statement X = Y + 3;":

To use a data item, PL/I must know the data type and how to process it. Attributes provide the information.

There are two kinds of attributes: data attributes and nondata attributes.

The following section describes computational data types and their attributes.

A source string is saved to a target string in a left-to-right order.

If a source string is longer than a target string, the excess part of the source string will be ignored and the STRINGSIZE condition will be triggered. If a target string is fixed-length and longer than a source string, the remaining part of the target string will be padded with the following characters:

If an arithmetic value of a source has the same attributes as a target except for its precision, a precision conversion must be performed.

For fixed-point data, point alignment is maintained during precision conversion. Therefore, padding or truncation can occur. For floating-point data, padding with 0 or truncation can occur.

Source-to-target rules are applied to conversions of data with the attributes listed below. For more information, refer to Source-to-Target Conversion Rules.

A value converted between decimal data and binary data can change. The conversion is performed using the factor 3.32 as follows:

The following describes how to convert each type:

A data item name is declared explicitly in the following cases. The scope of a data item declared explicitly is the block that includes the declaration.

The following shows an example scope of an explicit declaration. B and B' (C and C') indicate two different uses of the same name.

If a data item name is used in a program without explicit declaration, it is declared implicitly. The scope is the same as when a data item name is declared immediately after an external PROCEDURE statement in which the name is used. Some attributes of a data item declared implicitly are determined by context. This is called a contextual declaration and occurs in the following cases:

If an attribute is declared implicitly but not determined contextually, the default attribute is determined according to Defaults for Attributes. Since a contextual declaration cannot exist in the scope of an explicit declaration, it is impossible to add attributes by contextually evaluating the explicit declaration.

In most cases, the scope of a declared data item is determined by the position in which it is declared. The scope for an implicit declaration is the same as when a DECLARE statement is used immediately after the PROCEDURE statement of an external procedure.

The scope of a name that is explicitly declared in a block is the block. The name is unknown in outer blocks unless the same name is declared in the outer blocks. Even though the same name is declared in outer blocks, a data item that is different from the data item referred to by the inner block is referred to by the outer blocks. This characteristics allows a local definition in a program.

In the following example, the result for A in the PUT statement is 'FUNC1' because C.A is referred to and the result for B is 'TEST' declared in the TEST procedure.

The INTERNAL and EXTERNAL attributes define the scope of reference.

INTERNAL is the default attribute for entry names of internal procedures and all other variables except for entry constants, file constants, and user-defined conditions. It specifies that the scope of reference for a name is a block in which it is declared.

EXTERNAL is the default attribute for file constants, user-defined conditions, and entry constants, except for internal procedures. A name with this attribute can be declared multiple times in other external procedures. All declarations use the same name with matching EXTERNAL attribute and refer to the same data.

In the following example, if X.A in TEST1 is changed, X.A in TEST2 will be also changed. If X.B in TEST1 is changed, X.Y in TEST2 will be also changed.

Structure members always have the INTERNAL attribute.

Since EXTERNAL declarations for the same name refer to the same data, they must have the same attributes.

A computer accesses data in byte, halfword, fullword, or doubleword units. A halfword, fullword, and doubleword are 2, 4, and 8 consecutive bytes, respectively. Therefore, for halfword, fullword, and doubleword alignments, addresses with multiples of 2, 4, and 8 are accessed.

PL/I supports data alignment in a unit that a computer accesses. This allows unused bytes between successive data items, so that additional storage is required.

Whether to use data alignment in a unit is specified using the ALIGNED and UNALIGNED attributes.

Every name in a PL/I program requires a complete set of attributes. Arguments passed to a procedure must have attributes that match the procedure’s parameters. Values returned by a function must have expected attributes. However, user-specified attributes do not always require a complete set of attributes.

The following must use language-specified defaults or a new set of defaults determined by modifying language-specified defaults using a DEFAULT statement:

Attributes for explicitly or contextually declared names cannot be redefined by default attributes.

An array is an n-dimensional collection of elements with the same attributes. Each item of an array can be referred to based on the location in the array using a subscript. If an array is not specified using REFER, it must have at least one element for every dimension.

A structure is a collection of member elements that can be structures, arrays, and scalar variables. The name of a structure is used to refer to all data of the structure. Unlike an array, members of a structure can have different names and attributes. An asterisk (*) can be used to describe members that are not referred.

A structure has a level. A name described in level-1 is called a major structure. Names described under level-1 are called minor structures. Names described in the lowest level are called elementary variables.

A structure is declared by specifying a level-number (integer) before the name of a variable in a DECLARE statement.

The following example declares a structure.

If a member name of a structure is unique, the member can be referred to by using its name. If the same name is used by another member, the member cannot be referred to by only using its name. To refer to an exact member, a qualifier is required.

A qualifier is described by connecting parent member names using a point (.) in the order of levels. To refer to LAST in the above example, EMPLOYEE.LAST or EMPLOYEE.NAME.LAST can be used as a qualifier.

Since the scope of a variable is determined in a block unit, the variable Y referred to by the DISPLAY statement in the above example is checked from variables declared in the begin-block first.

Since only one Y is used in the begin-block, member variable Y of the structure X in the begin-block is referred to and 10 is outputted as the result value. If the variable Y is declared with the initialization value of 1 within the begin-block, the scalar variable Y, not the member variable Y of the structure X, is referred to and 1 is outputted as the result value.

In the above example, D is declared in two places. A.D refers to the second D, which is in the level number 5. To refer to the first D, A.C.D must be used as a qualifier.

In the above example, A.C cannot determine which C must be referred to.

In the above example, A, A.A, and A.A.A refer to the first, second, and third A, respectively.

In the above example, X refers to the first X. Y.Z cannot determine which Z must be referred to. Y.X.Z and Y.Y.Z refer to the first and second Z, respectively.

The LIKE attribute is used for a variable that is declared to use the same logical organization of another structure. The member names and their attributes of a target variable are declared in the same way as the members of a variable to declare. A target variable’s name and attributes are ignored.

A new member cannot be added to a structure.

The above declaration is the same as the following:

The variable A’s DIM(10), ALIGNED, and STATIC attributes are not affected by LIKE.

Storage is allocated when a block in which a variable is declared is activated, and released when the block is deactivated.

AUTOMATIC is the default attribute. Automatic variables are always internal.

The extent of an automatic variable can be described using restricted expressions that are expressions for which constant values are determined during compilation. Expressions for referring to another variable can also be used for this if the referred variable has an initial value when declared.

In the above example, the string length of the variable X is specified using the variable N. Since N has an initial value of 15, X becomes a string variable with the length of 15.

Storage is allocated when a program is loaded. It is not freed until the program execution completes.

STATIC is the default attribute for external variables and can be specified for internal variables. Variables declared outside a procedure have the STATIC attribute.

In the above example, the variable B has the STATIC attribute so that storage for B is allocated when the program execution starts. B can be referred to only within the block FUNC2 that includes the declaration of B. The variable A also has the STATIC attribute.

Static variables are initialized using the INITIAL attribute, and the initial variable must be a restricted expression. An array extent must also be specified using a restricted expression.

For variables declared with the CONTROLLED attribute, storage is allocated using an ALLOCATE statement. Storage is freed using a FREE statement or when a program execution ends. If a controlled variable for which storage is not allocated is referred to, an error will occur.

A non-controlled variable cannot be used to call a procedure that uses a controlled variable as a parameter.

The following example uses a controlled variable:

When an ALLOCATE statement is executed, storage for a controlled variable is allocated and stacked. Previously allocated and unreleased storage is pushed down. This stacking generates a new generation for the variable. The newly created generation becomes a current generation. The previous generation cannot be accessed unless the current generation is released. If a FREE statement is executed, the current generation storage is brought up and removed from the stack, and the previous generation becomes the current generation.

A based variable declaration is a description of a required storage size and its attributes. The location of the storage is identified using a locator. If an unallocated based variable is referred to, an error will occur.

Storage is allocated using the ALLOCATE statement and released using the FREE statement.

Based variables cannot have the EXTERNAL attribute. Any variable can be referred to using a locator.

If the REFER option is not used, the extent specifications in the BASED declarations must be restricted expressions with the following exceptions:

The INITIAL attribute sets an initial value to a variable when storage is allocated for the variable. This attribute cannot be used for defined data or parameters.

The following are examples of using iteration factors:

The following example initializes variables using the INITIAL attribute:

Z is an array that has 1,000 extents. Its 920 extents are initialized to 1 by the INITIAL attribute, and the remaining 80 extents are initialized according to repetitions of the sequence 5, 5, 5, and 9.

An AREA variable defines an area of storage allocated and released for a based variable. It can be declared using the AREA attribute and will have one of the 4 storage classes.

An area size can be specified using the same method for specifying string length and array size or the REFER option. It can also be specified as an asterisk (*).

The following example declares an AREA variable:

An area has an additional 16 bytes of memory for control information about current storage usage in addition to its defined size. Therefore, the sizes of AREA1 and AREA2 are 1,516 and 1,016 respectively.

When a based variable is allocated to an area, a part of the area is returned. When it is released, the area used by it becomes available for other variables. An area manages currently allocated storages using a chain. A head of a chain is included in the control information. Therefore, if control information is modified, an unexpected error can occur.

If the size of a based variable to allocate is less than a remaining area size, the AREA condition will occur.

The DEFINED attribute enables a part of storage or all of it for a base variable to be used as storage for a declared variable.

Defined variables do not inherit attributes from base variables. They are internal and must be level-1 variables. They can have the DIMENSION attribute, but cannot have the INITIAL, AUTOMATIC, BASED, CONTROLLED, STATIC, and PARAMETER attributes.

There are 3 types of defined variables:

The type is determined according to the following:

Simple defining can refer to other variables such as scalar, array, structure, and so on. The ALIGNED and UNALIGNED attributes of a defined variable and a base variable must be the same.

A is a three-dimensional array of X(2,2,2). B is a two-dimensional array of X(1,1,5) through X(5,5,5). The two defined variables are stored to unconnected storage, not adjacent storage.

C refers to the storage of X(L,M,N) that corresponds to the current values of L, M, and N.

Y is a string that has the first 10 characters of X.

Defined and base variables must be a string or picture data. They cannot have ALIGNED and VARYING attributes.

The POSITION attribute is used only for a string overlay. Base variables specify the starting position of defined variables.

X is a bit string that has 40 elements of B, starting from the 20th element.

X is a string array that consists of strings of the picture data A. Y is a string corresponding to the last 3 characters of A which are '999'. X is a string corresponding to the '9.99' value of A.

Y is a string that has the first 50 characters of the X string. POSITION(1) must be specified. Otherwise, it will have a simple definition which will produce a different result.

The OFFSET variable indicates the distance of an allocated based variable from the starting position of an area.

An OFFSET variable can be converted to a pointer value by being associated with an area if an AREA variable is specified. If an AREA variable is not specified, it must be specified explicitly using a POINTER built-in function.

A value can be saved to an OFFSET variable as follows: