PL/I data types, structures and built-in functions

PL/I describes data by attributes: whether a number is decimal or binary, fixed or floating, and how many digits it has. Strings can be fixed or varying, bits are a real type, and pointers let programs work with storage directly. A large library of built-in functions handles strings, arithmetic and dates.

Arithmetic data

A number in PL/I has a base (DECIMAL or BINARY), a scale (FIXED or FLOAT) and a precision. For business data you use fixed-point types almost exclusively.

Declaration	Stored as	COBOL equivalent	Typical use
FIXED DECIMAL(7,2)	Packed decimal, 4 bytes	PIC S9(5)V99 COMP-3	Money, quantities
FIXED BINARY(15)	Halfword (2 bytes)	PIC S9(4) COMP	Small counters
FIXED BINARY(31)	Fullword (4 bytes)	PIC S9(9) COMP	Counters, subscripts, lengths
FLOAT DECIMAL(16)	Floating point	COMP-2	Scientific values, not money

In FIXED DECIMAL(7,2) the first number is the total digits and the second the digits after the decimal point, so it holds up to 99999.99. This is the same packed decimal format COBOL uses, which is why the two languages can share files. See FIXED DECIMAL.

Strings and bits

CHARACTER(30) (or CHAR(30)) is always 30 bytes, padded with blanks. Equivalent to COBOL PIC X(30).
CHAR(30) VARYING holds up to 30 characters and stores the current length in a 2-byte prefix. COBOL has no direct equivalent, so mapping it needs care.
BIT(1) holds a single bit, written '1'B or '0'B. Ideal for flags such as end of file.

Structures and arrays

Structures use level numbers like COBOL, but with attributes in PL/I style. Arrays are declared by putting bounds after the name; the lower bound defaults to 1 but can be anything.

A customer structure and a table (illustrative)

 DCL 1 CUSTOMER,
       2 CUST_ID      CHAR(8),
       2 CUST_NAME    CHAR(30),
       2 BALANCE      FIXED DEC(11,2),
       2 ACTIVE       BIT(1);

 DCL MONTH_TOTAL(12)   FIXED DEC(13,2) INIT((12)0);
 DCL RATE_BAND(0:9)    FIXED DEC(5,4);

 CUSTOMER.BALANCE = CUSTOMER.BALANCE + 100.00;
 MONTH_TOTAL(3) = MONTH_TOTAL(3) + BALANCE;

INIT((12)0) uses a repeat factor to set all twelve entries to zero. Qualified names such as CUSTOMER.BALANCE use a dot, where COBOL writes BALANCE OF CUSTOMER.

Pointers and BASED storage

A POINTER holds an address. A variable declared BASED(P) has no storage of its own: it is a layout laid over whatever address P holds. This is how PL/I programs walk through buffers, chained lists and control blocks. It is also how they corrupt storage when a pointer is wrong. See BASED storage.

Pointer basicsWhat it means

DCL P POINTER;

A variable that holds an address.

DCL 1 HDR BASED(P), 2 LEN FIXED BIN(31), 2 TYPE CHAR(4);

A layout mapped at whatever P points to.

P = ADDR(BUFFER);

ADDR returns the address of a variable.

IF P = NULL() THEN ...

Check for a null pointer before using a based variable. Some code uses SYSNULL(); follow your site's convention.

Storage classes

AUTOMATIC (the default) is allocated when a block is entered and freed when it ends, so values do not survive between calls. STATIC is allocated once and keeps its value. BASED and CONTROLLED are allocated by the program. Uninitialised AUTOMATIC variables contain whatever was in storage, a classic source of intermittent bugs.

Built-in functions

Function	Purpose
SUBSTR(s, start, len)	Part of a string; can also be assigned to
LENGTH(s)	Current length of a string
INDEX(s, find)	Position of one string in another, 0 if absent
VERIFY(s, chars)	Position of the first character not in a set, 0 if all valid; useful for validating numeric input
TRANSLATE(s, to, from)	Replace characters, for example lower case to upper case
TRIM(s)	Remove leading and trailing blanks
MOD(a, b) / ROUND(x, n)	Remainder; rounding to n decimal places
HBOUND / LBOUND / DIM	Array upper bound, lower bound and extent
DATETIME()	Current date and time as a character string