Program Documentation: Program documentation falls into two
groups. Internal documentation which IS included as remarks within the
program and external documentation which gives a complete write-up of the
program. External documentation (frequently a manual) includes:
- overview write-up of what the program does
- layouts of the files, screens and reports
- detailed description of what the program does
- logic flowchart or pseudocode
- program printout (of the COBOL code)
- sample files, screens and outputs generated in test runs
- controls in the program
- operating description - when run, files needed, special paper etc.
- documentation of all modifications made to the program
SUMMARY:
- Understanding the project.
- Analyze the problem and develop a solution that is appropriate for
computer implementation.
- Develop the logic of the program.
- Write the program using a high level language and key it into the
computer using an text editor. The result is a SOURCE PROGRAM stored on
disk. This source program can be modified.
- COMPILE the SOURCE PROGRAM to produce an OBJECT PROGRAM. If syntax
errors are discovered during the compile, they must be fixed and the program
recompiled until an object program is produced.
- Link the OBJECT PROGRAM and produce an EXECUTABLE OBJECT PROGRAM.
- EXECUTE the EXECUTABLE OBJECT PROGRAM. If the results are correct, the
project is complete. If not. the programMER must reenter these steps at
step 1, 2, 3, or 4 depending on the complexity of the problem.
- DOCUMENT the program
Compiling
A COBOL program is written in an English type language that cannot be
understood by the machine, therefore it has to be translated using a
compiler. A compiler translates the code all at once and produces a
machine version of the program. (Contrast to an interpreter which is
supported by some other languages and translates a command and then
executes) The machine version is produced only when the program is free
of significant errors. TO DO A COMPILE: The compiler is
loaded into memory from the disk and it is the compiler that is being
executed:
| INPUT |
The SOURCE PROGRAM (high level language program) that has
been keyed in and stored as a file on a disk is read as input to
the compiler |
| PROCESSING |
The compiler attempts to translate the program. |
| OUTPUT |
- A listing of the program including any syntax errors that were
discovered or an on screen listing of the program with errors.
- If the compile was "clean", meaning no significant errors were
encountered, the OBJECT PROGRAM (machine language version of the
program) is produced and stored as a file on the disk.
|
If the attempt at a compile resulted in significant syntax errors (not
warnings), they must be corrected. The programmer will look at the list of
errors, determine the problem and determine how to correct it. Then the
programmer will bring up the text editor at the computer, bring up the
SOURCE PROGRAM file from the disk and use the text editor to make the
corrections. The corrected SOURCE PROGRAM will be saved and a new compile
will be tried. This process continues until a clean compile is achieved.
On some systems the link must be executed between the compile of the
program and the execution of the program. In other systems the link is
invisible and is done automatically. If your system requires the link step,
the following will be done. TO DO A LINK: The LINK program is loaded into
memory from the hard disk and then it is executed.
| INPUT |
Object program |
| PROCESSING |
The link resolves addresses, combines code etc.. |
| OUTPUT |
Executable OBJECT program |
Executing
Executing is running the executable object program that is a compiled and
linked translation of your COBOL program. When a program is executed, it
means that it is loaded into memory and the computer steps through the
instructions one instruction at a time and executes the instruction.
The program that the programmer wrote is now being executed.
After execution is complete, it is critical that the programmer carefully
check the output to make sure it is exactly what was intended. Frequently
test data that contains a lot of purposeful problems is used to test the
program to make sure that all programs are handled correctly. Usually, the
program does not work correctly the first time or for that matter, the first
couple of times that it is executed. Sometimes the program does nothing,
sometimes it process the first few records from the file it is reading and
then stops, sometimes the whole report it is producing is "garbage" other
times there are problems with a couple of columns on the report. Other
problems might be incorrect calculations, empty columns, missing or incorrect
totals etc. These errors are LOGIC ERRORS. It is the programmer's
job to examine what is happening and then go back to the program and
determine what is causing the problem. Sometimes the problem is easy to
solve. For example the missing column might mean the programmer simply
forgot to move anything to that column and the missing totals might simply
mean the programmer forgot to write the totals on the report. Other times
the problem is complex and requires going back and analyzing the problem,
rethinking some of the logic and making significant modifications to the
instructions (code) of the program.
The systems analyst or someone else in charge of the project should also
carefully review the output to make sure it is what is needed. This is very
important, because sometimes a "second set of eyes" will catch something
that the programmer missed! If the program works correctly it is usually
tested more completely on live data and system tested which means output
from one program is read as input to another program to make sure that data
is being processed correctly (note your program could be the one producing
the output or reading the input). When the program has successfully passed
all of its tests, it is put into production.
If the program does not work correctly, the programmer has to figure out
what is wrong, go back to the text editor and bring up the source program,
fix it, recompile it (hopefully without errors, but if not they will have to
be fixed), and then test the program again to see if the desired output has
been produced. Honest, programming is fun!!!
TO EXECUTE A PROGRAM: The executable program is loaded into
memory from the disk.
| INPUT |
The input to this process is whatever input is called for by the
program. It may involve reading one or more disk files, accepting
data from the keyboard etc. |
| PROCESSING |
The executable object program is executed one instruction at a time
until the program is complete (machine translation of STOP RUN is
encountered) |
| OUTPUT |
The output from this process is whatever output the program
generates. It might be a report, it might be displays on the screen,
it might be one or more disk files etc.
|
Programming Logic
Program Design:
Program design is a multiple step process. One approach that works well is:
- Analyze the problem to make you sure you understand what
your program is supposed to accomplish. Talk to the systems analyst and/or
the program requestor for clarification.
- Analyze the data. The programmer needs to look at the
specifications for the output the program is supposed to produce and look at
the input files or specifications and truly understand the layout. If the
data being read is on existing files, the programmer should throughly
understand what is contained in each field within the file. This includes
understanding what is numeric, what is alphanumeric, what is coded, the
length of the data, whether it is packed, display or binary, whether it is
signed, whether it is grouped and whether the progammer needs to access only
certain parts of the data in a field. The programmer should understand what
data is valid in each field, whether the data that the program will read is
edited data or whether invalid data needs to be screened. Sometimes the
input data can be used directly to create the output, sometimes calculations
need to be done or decisions need to be made and sometimes after analyzing
the data it is clear that the data is insufficient to produce the desired
output.
- Analyze the processing to determine exactly what needs to
be done to produce the desired output from the given input. This process is
called FUNCTIONAL DECOMPOSITION (explained in the next section).
- Design the logic of the program. This means layout out a
step by step plan using a logic flowchart, pseudocode or some other logic
tool that the programmer finds of value.
Functional Decomposition:
Functional decomposition is breaking down a large
task/function/program into smaller parts which are known as subfunctions.
To start the process of functional decomposition, the programmer must
indentify the inputs, outputs and major tasks to be accomplished by the
program. Next, the programmer logically groups the required tasks as
functions. In doing functional decomposition, the programmer focuses on
the tasks that are necessary to turn the given input into the required
output. The focus here is on the process that will have to be accomplished,
not how many times you will do the task.
Therefore, the starting point in functional decomposition is to examine the
output and determine what is needed. Then the programmer looks at the input
to determine if the data includes everything that will be needed to produce
the output. Finally the programmer, decides on the processing that is
necessary to turn the given input into the required output. The questions
to ask are:
- Exactly what output is required?
- What input is needed to produce the required output?
- What processing is needed to turn the given input into the required
output?
Once the necessary functions have been defined, the programmer must break
the functions down into activities. A function frequently includes multiple
interrelated activites. For example, if the function is to calculate net
pay then the activities might include: checking to see if the person is
salary or hourly and if they are hourly calculating gross pay, then
subtracting deductions and taxes from gross pay to get net pay. Each of
these decisions and mathematical calculations is an activity within the
calculate net pay function. Functional decomposition is the essence behind
the concept of top-down design.
Top-down design is starting with the whole problem and then breaking the
problem down into a series of tasks or functions which can then be broken
down into detailed activities,essentially doing functional decomposition.
Frequently the resulting break down is represented using a HIERARCHY CHART.
Obviously, the next step is to take the functional decomposition or top-down
design and proceed to the programming stage.
TOP-DOWN PROGRAMMING
Top-down programming is very compatible with structured programming.
Essentially, top-down programming calls for the program to be broken up into
modules with the high-level modules containing the programming elements that
control the program and the low-level modules containing the details. In
the sample programs, the program is first broken up into three modules:
A-100-INITIALIZE, B-100-PROCESS, and C-100-TERMINATE. These modules will
determine when the next level down, the 200 level will be executed, and in
turn the 200 level will determine when the 300 level will be executed. For
example B-100-PROCESS performs B-200-LOOP and B-200-LOOP determines when to
perform B-300-HDR-ROUT. In this series of modules, the B-100-PROCESS
contains the big picture and controls the entire processing element of the
program. The B-200-LOOP controls how each individual record is processed.
The details may appear in the B-200-LOOP or if the coding is too complex,
the B-200-LOOP may have a command to perform B-310-CALC where the
calculations are done or B-320-SETUP where the printline is setup. If a
report is being produce, at the least, the B-200-LOOP will probably have the
command to perform B-300-HDR-ROUT where the headers for each page are done.
When writing a program the functions are grouped into three categories:
- Initialization
- Processing
- Termination
The mainline (MAINLINE) of the program is the first level of your hierarchy
chart.
- INITIALIZING
- open the files
- get the date if needed
- PROCESSING
- read the initializing/first record
- set any hold areas or indicators, if needed
- perform a processing loop (B-200-LOOP) until a certain condition is met -
frequently the condition is end of file
- the processing loop - B-200-LOOP includes the instructions to process
the record, write it and then read another record - it may execute other
paragraphs such as a B-300-ROUT and if needed a B-310-CALC-ROUT
- after processing is the B-200-LOOP is complete, do any wrapup processing
such as writing final total lines in a B-210-FINAL-TOTALS
- TERMINATION
The structure of these modules is shown in a HIERARCHY CHART. The hierarchy
chart shows the modules by name, but does not give any information about the
processing that takes place in the module. Its function is to show the
control, which module executes which module and where the control will
return after the module has been executed. To understand the detail in each
module, the programmer can check the flowchart, the pseudocode or the actual
COBOL code.
Structured Programming
Structured programming is a method of programming that helps the programmer
develop clean, well written programs. It has been shown that a structured
program is more readable, easier to test and debug, easier to maintain, more
reliable, and more effective. Structured programming is comprised of three
logical structures: sequence, selection and iteration.
Sequence structure: In a sequential structure, the commands
are executed in sequence. The flow of the program is to complete one
instruction and then drop down and execute the next instruction and then the
next until something terminates the sequence such as the end of a paragraph.
Example:
MOVE NAME-IN TO NAME-PR.
MULTIPLY AMT-IN BY 1.5
GIVING AMT-PLUS-WS.
MOVE AMT-PLUS-WS TO AMT-PLUS-PR.
WRITE PRINTZ
AFTER ADVANCING 1 LINES.
Selection structure: In a selection structure the processing
is dependent on a condition that is being tested. In COBOL, the selection
structure is usually accomplished with an IF or an EVALUATE (the
implementation of the case structure in COBOL) or with an implied IF such
as the AT END clause in the READ statement.
Example #1:
IF EMPLOYEE-CODE = "S"
MOVE SALARY TO SALARY-PR
ELSE
MULTIPLY PAY-HRS BY HRS-WK
GIVING WEEK-PAY-WS
MOVE WEEK-PAY-WS TO WEEK-PAY-PR.
Note: The ELSE in this example also uses the sequence structure since there
are two commands - this example indicates the combination of the two
structures.
Example #2: (implied if)
READ INPUT-FILE
AT END
MOVE "YES" TO EOF-IND.
Iteration structure: (LOOP STRUCTURE) The iteration structure
causes something to be executed over and over again until some condition
terminates the repetition. This structure is essentially the looping
structure that has been used in all of the programs. When defining
iteration, there are two basic structures that a language may implement:
Do-While and Do-Until. The difference between the two structures is when
the condition is tested. In the Do-While structure the condition is tested
before the loop is executed while in the Do-Until structure the condition is
tested after the loop has been executed. This means that with the Do-While
structure there is a possibility that the loop will never be executed. The
PERFORM...UNTIL used in the sample programs is an example of the Do-While
structure because the condition is tested before the loop is executed. For
example in the PERFORM B-200-LOOP UNTIL EOF-IND = "YES", if the EOF-IND is
already set to YES the B-200-LOOP will never be executed. In the Do-Until
structure, since the test is done after the loop has been executed the loop
will always be executed at least once. COBOL allows the programmer to modify
the PERFORM...UNTIL to conform to the Do-Until structure by adding the
clause WITH TEST AFTER.
Example of COBOL PERFORM using the standard DO-WHILE structure:
PERFORM B-200-LOOP
UNTIL EOF-IND = "YES".
Example of COBOL PERFORM using the DO-UNTIL structure:
PERFORM B-300-CALC
WITH TEST AFTER
UNTIL CT-FLD > 12.
In this example, B-300-CALC will be executed before the CT-FLD > 12 test is
done, therefore it will always be executed once. After B-300-CALC has been
executed once, the test will be made to determine if the paragraph should be
executed again.