2. Basic terminology

This chapter introduces the words the rest of the manual leans on (§2). They form a small vocabulary for talking about the parts of a FORTRAN program. None of it is code you write; it is the scaffolding the standard uses to describe what you write.

Programs and program units

An executable program is a self-contained computing procedure (§9.1.6). It consists of precisely one main program and, optionally, one or more subprograms.

• A main program is a set of statements and comments that does not begin with a FUNCTION, SUBROUTINE, or BLOCK DATA statement (§9.1.5). Every executable program has exactly one.
• A subprogram is a unit that is headed by one of those statements:
  • one headed by FUNCTION or SUBROUTINE is a procedure subprogram — it specifies actions (a computation, a side effect);
  • one headed by BLOCK DATA is a specification subprogram — it only supplies initial values to named common, and specifies nothing to execute.

The umbrella term program unit means either a main program or a subprogram. A program unit that another unit can invoke is an external procedure: a FUNCTION subprogram is referenced as a function (it returns a value into an expression), a SUBROUTINE subprogram is invoked by a CALL statement.

      PROGRAM MAIN
      X = SQ(3.0)
C     -> references the function SQ
      CALL SHOW(X)
C     -> invokes the subroutine SHOW; prints "  9.00"
      STOP
      END
C
C     a procedure subprogram (function)
      FUNCTION SQ(A)
      SQ = A*A
      RETURN
      END
C
C     a procedure subprogram (subroutine)
      SUBROUTINE SHOW(V)
      WRITE(6,100) V
100   FORMAT(1X,F6.2)
      RETURN
      END

Chapters 8 and 9 cover procedures and the make-up of a complete program in full; here we only need the names.

Lines, statements, and comments

A program unit is built from lines. A statement is written on one or more lines: the first is the initial line and any others are continuation lines (§3.2). A third kind of line, a comment line, is not part of any statement — it exists purely to document the program for a human reader.

Statements fall into two broad classes (§7.1, §7.2):

• executable statements specify action — an assignment, a jump, a CALL, an I/O operation;
• nonexecutable statements describe the program rather than act: they declare the characteristics of data (DIMENSION, COMMON, type-statements), supply editing information (FORMAT), define statement functions, or arrange data (DATA).

The exact column layout that distinguishes an initial line from a continuation or a comment is the subject of Chapter 3.

Names, operators, and lists

The syntactic pieces of a statement are names and operators (§5). Names identify the things a program works on — data (variables, arrays, constants) and procedures. Operators (+, *, .AND., and the imperative verbs like GO TO) specify the action taken on what the names identify.

One kind of name is worth singling out now: an array name. An array is a named, ordered collection of data declared in an array declarator (§7.2.1.1); a single element is identified by qualifying the array name with a subscript (§5.1.3), as in A(2,3). Arrays are the subject of much of Chapter 7.

Throughout the standard (and this manual) a list is a sequence of items separated by commas — a subscript list (I,J), an I/O list, a COMMON list. The standard notes that such a list must contain at least one item unless an exception is stated (§2).

forterp notes. forterp builds an executable program from one or more source files: you can keep the main program and each subprogram in separate files and forterp links them, the way a classic LINK/LOADER step would (see Command-line tools for multi-file runs). The main program is identified structurally — the unit that is not headed by FUNCTION, SUBROUTINE, or BLOCK DATA — exactly as in §9.1.5, so a leading PROGRAM statement is convenient but not required for forterp to find it.