[Skip Navigation] [CSUSB] / [CNS] / [Comp Sci Dept] / [R J Botting] / [Samples] / A
[Index] [Contents] [Source Text] [About] [Notation] [Copyright] [Comment/Contact] [Search ]
Tue Sep 18 15:25:03 PDT 2007

Contents


    Language Reference Manual for the Language A

      Context

      This document is an example and template for a language reference manual --(LRM). It defines a very simple programming language called A.

      The document defines the language piece by piece. For each part it defines the syntax and semantics. It also comments on the purpose for the part and gives examples of the code described. Good LRMs need examples, comments, syntax, and semantics.

      This manual was written using XBNF and translated to HTML by Dr. Botting's mth2html translator that also indexed and added the contents list and inserted hyperlinks. These are also parts of good manuals.

      This is an example for a class and is the basis for project work, so I've not documented some of the options and issues that arise in the design languages. I've deliberately added one feature that is a bad idea. I wonder if you can spot the language that I ripped it from.... and added a perverse twist of my own

      Warning

      This is a bad language. Do not try it at home. If you feel like wtiing a compiler -- it is not difficult BUT its better to fix the problems with it FIRST.

      The Language A

        A is not unlike half-a-dozen "autocodes" used in the 1960's in the United Kingdom. Real languages in this family include: Pegasus Autocode, Elliott 803 Autocode, Mercury Autocode, and K-code. All were compiled, targeted scientific computing, and are not mentioned in most text books. They did not survive competition from FORTRAN, COBOL, and Algol. However, they are small enough to be defined by a small LRM. A is slightly more complex than Pegasus Autocode, and definitely less complex than K-code. The big omission is any form data structure, and so, no indexing or subscripts.

        Programs

        An A program consists of a series of executable statements (each on one line), and terminated by an "END" statement:
      1. program::= #(statement eoln) "END" eoln.
      2. eoln::=end of line, any sequence of white space characters including at least one "Carriage Return" character.

        Here is an example of an A program:

         READ a
         b=a*a
         PRINT b
         END
        The above program reads a floating point decimal number (7.0 for example, and puts it into floating point location a. The next squares it, and stores the result in b. Then it prints out the result(49.0).

        A program in A starts by executing the the first statement and finishes when the END statement is executed. Normally each statement is taken in sequence, however, some statements can select a different successor statement that is before or after them in the sequence. For example to print the sum of the first n integers one writes:

         i=1
         s=0
         READ n
         j=i>=n
         SKIP 4*j
         s=s+i
         i=i+1
         SKIP -4
         PRINT s
         END

        The following program does the same thing using a formula (s = n(n+1)/2) instead of a loop.

         i=1
         READ n
         s=n+1
         s=s*n
         s=s/2
         PRINT s
         END
        Notice how complex formaluae are written as a series of simple ones.

        Statements

        There are four types of statement (other than the END statement):
      3. statement::= assignment | input | output | skip.

      4. assignment::= variable "=" expression.
         d=b*b
         e=4*a
         e=e*c
         d=d-e

      5. input::= "READ" variable.
         READ a
         READ n

      6. output::="PRINT" expression | "OUT" text.
         PRINT d
         OUT Real Roots

      7. skip::="SKIP" expression.
         SKIP -12
         SKIP +3
         SKIP 0
         SKIP +17*k
        The following indicates which statement is executed given the values 0, -1, -2, +1, and +2
         (-2) ...
         (-1)...
             SKIP ?
         (0) ....
         (+1) ...
         (+2) ...

        Variables

        Variable are represented by the 26 lowercase letters. Letters i through r inclusive are for integer (fixed point ) numbers. The rest are hold floating point numbers.
      8. variable::="a" .. "z".

        Expressions

        An expression has at most one operator and one sign or a single function call.
      9. expression::=O(sign) ( part O( operator second_part ) | function part ).

      10. part::= variable | constant.
      11. second_part::=part.
      12. sign::= "+" | "-".
      13. operator::= sign | "*" | "/" | relation.
      14. relation::="<" | ">" | "<=" | ">=" | "<>" | "==".
      15. function::="SQRT" | "ABS" | "LOG" | "COS" | "SIN" | "TAN" | "EXP".

        Examples

         x
         -x
         i+1
         x+i
         x+1.234
         -x+y
         +x*y
         -3

        Relations

        The values of a relation are always either one(true) or zero(false). They are typically used in SKIP statements to choose different successors depending on the values of variable. For example the following code places the larger of x and y into z.
         j=x>y
         SKIP 2*j
         z=y
         SKIP 1
         z=x
        exercise: can you write code to put the minimum of x, y into z?

        Constants

        A constant is a number: fixed point integer or floating point number with a decimal point. Only decimal notation is used.
      16. constant::= fixed | floating.
      17. fixed::= N(digit).
         1234

      18. floating::= N(digit) "." N(digit).

         12.34
        Notice that "1." and ".95" must be written as "1.0" and "0.95" respectively.

        Longer Example: Solving a Quadratic (Untested)

         READ a
         READ b
         READ c
         d=b*b
         e=4*a
         e=e*c
         d=d-e
         e=2*a
         r=d<0
         SKIP r*7
         d=SQRT d
         s=-b+d
         s=s/e
         t=-t-d
         t=t/e
         PRINT s
         PRINT t
         SKIP 7
         d=d-d
         d=SQRT d
         s=-b/e
         t=d/e
         PRINT s
         OUT + or - i *
         PRINT t
         END
        Exercise: handle the case when a is zero.

        Semantics

        Here is an outline of the semantic structure of the language A in the UML. [UML semantic overview]

        Note

        A is not the smallest possible programming language.

      . . . . . . . . . ( end of section The Language A) <<Contents | End>>

      Notation

    1. For X, N(X)::=1 or more X's,
    2. For X, O(X)::=optional X, 0..1, [ cs320xbnf.htm ]

    3. digit::="0".."9", [ cs320xbnf.htm ]

    . . . . . . . . . ( end of section Language Reference Manual for the Language A) <<Contents | End>>

End