Testing is a critical aspect of COBOL application development and maintenance. Thorough testing ensures that programs function correctly, meet business requirements, and maintain reliability even after modifications. For COBOL applications that often support mission-critical business functions, effective testing is particularly important.
| Challenge | Impact | Mitigation Strategies |
|---|---|---|
| Complex Legacy Codebases | Difficult to isolate components for testing | Incremental refactoring, modularization |
| Limited Testing Tools | Fewer automated testing options | Specialized mainframe testing tools, custom frameworks |
| Environment Constraints | Limited MIPS, batch windows, resource contention | Test environment optimization, scheduling |
| Knowledge Gaps | Retiring workforce, undocumented logic | Knowledge transfer, code analysis tools |
| Integration Complexity | Interdependencies with other systems | Service virtualization, stubs/mocks |
A comprehensive testing strategy for COBOL applications should include all these types of testing, adapted to the specific requirements and constraints of the mainframe environment.
Unit testing involves testing individual components of a program in isolation to verify that each part functions correctly on its own. While unit testing can be challenging for legacy COBOL applications, several approaches have proven effective.
Test drivers are small COBOL programs written specifically to test a component by providing controlled inputs and verifying outputs.
12345678910111213141516171819202122232425262728293031323334353637383940414243IDENTIFICATION DIVISION. PROGRAM-ID. TEST-DRIVER. DATA DIVISION. WORKING-STORAGE SECTION. 01 TEST-INPUTS. 05 INPUT-VALUE-1 PIC 9(5)V99 VALUE 1234.56. 05 INPUT-VALUE-2 PIC 9(5)V99 VALUE 7890.12. 05 EXPECTED-RESULT PIC 9(6)V99 VALUE 9124.68. 01 TEST-OUTPUTS. 05 ACTUAL-RESULT PIC 9(6)V99. 05 TEST-PASSED-FLAG PIC X. 88 TEST-PASSED VALUE 'Y'. 88 TEST-FAILED VALUE 'N'. PROCEDURE DIVISION. MAIN-PARAGRAPH. DISPLAY "BEGINNING TEST OF CALC-SUM PARAGRAPH" * Set up test inputs MOVE 1234.56 TO INPUT-VALUE-1 MOVE 7890.12 TO INPUT-VALUE-2 * Call the paragraph being tested PERFORM CALC-SUM * Verify results IF ACTUAL-RESULT = EXPECTED-RESULT SET TEST-PASSED TO TRUE DISPLAY "TEST PASSED: Result = " ACTUAL-RESULT ELSE SET TEST-FAILED TO TRUE DISPLAY "TEST FAILED:" DISPLAY " Expected: " EXPECTED-RESULT DISPLAY " Actual: " ACTUAL-RESULT END-IF STOP RUN. * This is the paragraph being tested CALC-SUM. COMPUTE ACTUAL-RESULT = INPUT-VALUE-1 + INPUT-VALUE-2 .
Test drivers can be used to test individual paragraphs, sections, or entire subprograms. They are particularly useful for testing complex business logic.
Stubs and mocks replace external dependencies with simplified versions that return predefined data, allowing you to isolate the component being tested.
12345678910111213141516171819202122232425262728293031323334353637383940414243444546* Original code with external dependency CUSTOMER-LOOKUP. MOVE CUST-ID-IN TO DB-CUSTOMER-ID CALL 'DB-PROGRAM' USING DB-REQUEST-AREA IF DB-STATUS-OK MOVE DB-CUSTOMER-NAME TO CUST-NAME-OUT MOVE DB-CUSTOMER-ADDR TO CUST-ADDR-OUT ELSE MOVE SPACES TO CUST-NAME-OUT MOVE SPACES TO CUST-ADDR-OUT END-IF. * Test stub replacing the database call IDENTIFICATION DIVISION. PROGRAM-ID. DB-PROGRAM. DATA DIVISION. LINKAGE SECTION. 01 DB-REQUEST-AREA. 05 DB-FUNCTION PIC X(8). 05 DB-CUSTOMER-ID PIC X(10). 05 DB-STATUS PIC XX. 88 DB-STATUS-OK VALUE "00". 05 DB-CUSTOMER-DATA. 10 DB-CUSTOMER-NAME PIC X(30). 10 DB-CUSTOMER-ADDR PIC X(50). PROCEDURE DIVISION USING DB-REQUEST-AREA. MAIN-PARAGRAPH. * Simulate database lookup with test data EVALUATE DB-CUSTOMER-ID WHEN "1234567890" MOVE "00" TO DB-STATUS MOVE "JOHN DOE" TO DB-CUSTOMER-NAME MOVE "123 MAIN ST" TO DB-CUSTOMER-ADDR WHEN "9876543210" MOVE "00" TO DB-STATUS MOVE "JANE SMITH" TO DB-CUSTOMER-NAME MOVE "456 OAK AVE" TO DB-CUSTOMER-ADDR WHEN OTHER MOVE "01" TO DB-STATUS MOVE SPACES TO DB-CUSTOMER-NAME MOVE SPACES TO DB-CUSTOMER-ADDR END-EVALUATE GOBACK.
JCL (Job Control Language) can be used to set up automated unit tests for batch COBOL programs.
1234567891011121314151617181920212223242526//UNITTEST JOB (ACCT),'UNIT TEST',CLASS=A,MSGCLASS=X //* //* JCL TO EXECUTE UNIT TESTS FOR THE CUSTOMER PROCESSING PROGRAM //* //STEP01 EXEC PGM=CUSTPROC //STEPLIB DD DSN=TEST.LOADLIB,DISP=SHR //* //* TEST INPUT DATA //CUSTINP DD DSN=TEST.DATA.CUSTTEST,DISP=SHR //* //* OUTPUT WILL BE COMPARED TO EXPECTED RESULTS //CUSTOUT DD DSN=TEST.DATA.ACTUAL,DISP=(NEW,CATLG), // SPACE=(CYL,(1,1)), // DCB=(RECFM=FB,LRECL=100,BLKSIZE=0) //* //SYSOUT DD SYSOUT=* //SYSUDUMP DD SYSOUT=* //* //* COMPARE ACTUAL OUTPUT TO EXPECTED OUTPUT //STEP02 EXEC PGM=IDCAMS //SYSPRINT DD SYSOUT=* //SYSIN DD * COMPARE - DATA(TEST.DATA.ACTUAL) - WITH(TEST.DATA.EXPECTED) /*
This approach uses standard mainframe utilities like IDCAMS COMPARE to verify that the program produces the expected output files given known input data.
Modern COBOL unit testing can leverage specialized frameworks:
1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465* Example using a hypothetical unit testing framework IDENTIFICATION DIVISION. PROGRAM-ID. CALC-INTEREST-TEST. ENVIRONMENT DIVISION. CONFIGURATION SECTION. SPECIAL-NAMES. TEST-FRAMEWORK IS UNITTEST. DATA DIVISION. WORKING-STORAGE SECTION. 01 UNITTEST-FRAMEWORK. 05 TEST-CASE-NAME PIC X(30). 05 TEST-RESULT PIC X. 88 TEST-PASSED VALUE 'P'. 88 TEST-FAILED VALUE 'F'. 05 TEST-MESSAGE PIC X(100). 01 TEST-DATA. 05 PRINCIPAL PIC 9(7)V99. 05 RATE PIC 9(2)V9(6). 05 TERM PIC 9(2). 05 EXPECTED-INTEREST PIC 9(7)V99. 05 ACTUAL-INTEREST PIC 9(7)V99. PROCEDURE DIVISION. TEST-MAIN. PERFORM SETUP-TEST-CASE PERFORM RUN-TEST-CASES UNITTEST FINISH-TESTS. STOP RUN. SETUP-TEST-CASE. UNITTEST INITIALIZE. RUN-TEST-CASES. * Test case 1: Simple interest calculation MOVE "CALC-SIMPLE-INTEREST-TEST-1" TO TEST-CASE-NAME UNITTEST START-TEST USING TEST-CASE-NAME MOVE 1000.00 TO PRINCIPAL MOVE 0.05 TO RATE MOVE 1 TO TERM MOVE 50.00 TO EXPECTED-INTEREST CALL 'CALC-INTEREST' USING PRINCIPAL, RATE, TERM, ACTUAL-INTEREST UNITTEST ASSERT-EQUALS EXPECTED-INTEREST ACTUAL-INTEREST UNITTEST END-TEST. * Test case 2: Zero principal MOVE "CALC-SIMPLE-INTEREST-TEST-2" TO TEST-CASE-NAME UNITTEST START-TEST USING TEST-CASE-NAME MOVE 0 TO PRINCIPAL MOVE 0.05 TO RATE MOVE 1 TO TERM MOVE 0 TO EXPECTED-INTEREST CALL 'CALC-INTEREST' USING PRINCIPAL, RATE, TERM, ACTUAL-INTEREST UNITTEST ASSERT-EQUALS EXPECTED-INTEREST ACTUAL-INTEREST UNITTEST END-TEST.
Effective testing requires appropriate test data that covers various scenarios, boundary conditions, and error cases. Creating comprehensive test data is particularly important for COBOL applications that often process large volumes of records with complex business rules.
Well-designed test data should meet these criteria:
For smaller tests or specialized cases, manually created test data can be effective:
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103* Example of a COBOL program to generate test data IDENTIFICATION DIVISION. PROGRAM-ID. TESTGEN. ENVIRONMENT DIVISION. INPUT-OUTPUT SECTION. FILE-CONTROL. SELECT TEST-FILE ASSIGN TO TESTOUT ORGANIZATION IS SEQUENTIAL. DATA DIVISION. FILE SECTION. FD TEST-FILE RECORDING MODE IS F RECORD CONTAINS 100 CHARACTERS. 01 TEST-RECORD. 05 CUSTOMER-ID PIC X(10). 05 CUSTOMER-NAME PIC X(30). 05 TRANSACTION-AMT PIC 9(9)V99. 05 TRANSACTION-DATE PIC X(8). 05 TRANSACTION-TYPE PIC X. 05 FILLER PIC X(40). WORKING-STORAGE SECTION. 01 COUNTERS. 05 RECORD-COUNT PIC 9(5) VALUE ZERO. 05 IDX PIC 9(5) VALUE ZERO. 01 TEST-DATA-VALUES. 05 TRANS-TYPES PIC X(5) VALUE "CDPRA". 05 TRANS-TYPE-COUNT PIC 9 VALUE 5. PROCEDURE DIVISION. MAIN-PARAGRAPH. OPEN OUTPUT TEST-FILE * Generate standard test cases PERFORM GENERATE-NORMAL-CASES * Generate boundary condition test cases PERFORM GENERATE-BOUNDARY-CASES * Generate error condition test cases PERFORM GENERATE-ERROR-CASES CLOSE TEST-FILE DISPLAY "Generated " RECORD-COUNT " test records" STOP RUN. GENERATE-NORMAL-CASES. * Regular transaction with positive amount MOVE "0000000001" TO CUSTOMER-ID MOVE "JOHN SMITH" TO CUSTOMER-NAME MOVE 1250.95 TO TRANSACTION-AMT MOVE "20230615" TO TRANSACTION-DATE MOVE "C" TO TRANSACTION-TYPE PERFORM WRITE-TEST-RECORD * Regular transaction with zero amount MOVE "0000000002" TO CUSTOMER-ID MOVE "JANE DOE" TO CUSTOMER-NAME MOVE ZERO TO TRANSACTION-AMT MOVE "20230616" TO TRANSACTION-DATE MOVE "P" TO TRANSACTION-TYPE PERFORM WRITE-TEST-RECORD. GENERATE-BOUNDARY-CASES. * Maximum transaction amount MOVE "0000000003" TO CUSTOMER-ID MOVE "BOUNDARY TEST MAX" TO CUSTOMER-NAME MOVE 999999999.99 TO TRANSACTION-AMT MOVE "20230617" TO TRANSACTION-DATE MOVE "D" TO TRANSACTION-TYPE PERFORM WRITE-TEST-RECORD * Minimum valid date MOVE "0000000004" TO CUSTOMER-ID MOVE "BOUNDARY TEST MIN DATE" TO CUSTOMER-NAME MOVE 100.00 TO TRANSACTION-AMT MOVE "19000101" TO TRANSACTION-DATE MOVE "R" TO TRANSACTION-TYPE PERFORM WRITE-TEST-RECORD. GENERATE-ERROR-CASES. * Invalid transaction type MOVE "0000000005" TO CUSTOMER-ID MOVE "ERROR TEST INVALID TYPE" TO CUSTOMER-NAME MOVE 100.00 TO TRANSACTION-AMT MOVE "20230618" TO TRANSACTION-DATE MOVE "X" TO TRANSACTION-TYPE PERFORM WRITE-TEST-RECORD * Invalid date format MOVE "0000000006" TO CUSTOMER-ID MOVE "ERROR TEST INVALID DATE" TO CUSTOMER-NAME MOVE 100.00 TO TRANSACTION-AMT MOVE "20231350" TO TRANSACTION-DATE MOVE "C" TO TRANSACTION-TYPE PERFORM WRITE-TEST-RECORD. WRITE-TEST-RECORD. WRITE TEST-RECORD ADD 1 TO RECORD-COUNT.
Mainframe utilities can be used to generate or manipulate test data efficiently:
123456789101112131415161718192021222324252627282930313233343536//GENSORT JOB (ACCT),'TEST DATA GEN',CLASS=A,MSGCLASS=X //STEP1 EXEC PGM=SORT //SORTIN DD DUMMY //SORTOUT DD DSN=TEST.CUSTOMER.DATA, // DISP=(NEW,CATLG), // SPACE=(CYL,(1,1)), // DCB=(RECFM=FB,LRECL=100,BLKSIZE=0) //SYSIN DD * OPTION COPY OUTFIL REMOVECC, BLKCCH=100, IFTHEN=(WHEN=INIT, BUILD=(1:C'0000000001', 11:C'JOHN SMITH ', 41:C'123 MAIN ST ', 71:C'19760523', 79:C'A', 80:C'0000100000', 90:C' ')), IFTHEN=(WHEN=INIT, BUILD=(1:C'0000000002', 11:C'JANE DOE ', 41:C'456 OAK AVE ', 71:C'19820617', 79:C'B', 80:C'0000250000', 90:C' ')), IFTHEN=(WHEN=INIT, BUILD=(1:C'0000000003', 11:C'ALICE JOHNSON ', 41:C'789 PINE BLVD ', 71:C'19900712', 79:C'C', 80:C'0000075000', 90:C' ')) /*
This JCL example uses DFSORT to generate test customer records with different values.
Using production data for testing can provide realistic scenarios but requires careful handling:
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101* Example of a simple data masking program IDENTIFICATION DIVISION. PROGRAM-ID. DATAMASK. ENVIRONMENT DIVISION. INPUT-OUTPUT SECTION. FILE-CONTROL. SELECT PROD-FILE ASSIGN TO PRODIN ORGANIZATION IS SEQUENTIAL. SELECT MASKED-FILE ASSIGN TO MASKOUT ORGANIZATION IS SEQUENTIAL. DATA DIVISION. FILE SECTION. FD PROD-FILE RECORDING MODE IS F RECORD CONTAINS 100 CHARACTERS. 01 PROD-RECORD. 05 PR-CUSTOMER-ID PIC X(10). 05 PR-CUSTOMER-NAME PIC X(30). 05 PR-SSN PIC X(9). 05 PR-BIRTH-DATE PIC X(8). 05 PR-ACCOUNT-NUM PIC X(16). 05 PR-BALANCE PIC 9(9)V99. 05 PR-FILLER PIC X(14). FD MASKED-FILE RECORDING MODE IS F RECORD CONTAINS 100 CHARACTERS. 01 MASKED-RECORD. 05 MR-CUSTOMER-ID PIC X(10). 05 MR-CUSTOMER-NAME PIC X(30). 05 MR-SSN PIC X(9). 05 MR-BIRTH-DATE PIC X(8). 05 MR-ACCOUNT-NUM PIC X(16). 05 MR-BALANCE PIC 9(9)V99. 05 MR-FILLER PIC X(14). WORKING-STORAGE SECTION. 01 FILE-STATUS. 05 PROD-EOF PIC X VALUE 'N'. 88 END-OF-PROD-FILE VALUE 'Y'. PROCEDURE DIVISION. MAIN-PARAGRAPH. OPEN INPUT PROD-FILE OPEN OUTPUT MASKED-FILE PERFORM UNTIL END-OF-PROD-FILE READ PROD-FILE AT END SET END-OF-PROD-FILE TO TRUE NOT AT END PERFORM PROCESS-RECORD END-READ END-PERFORM CLOSE PROD-FILE CLOSE MASKED-FILE STOP RUN. PROCESS-RECORD. * Copy the record structure MOVE PROD-RECORD TO MASKED-RECORD * Mask sensitive data PERFORM MASK-SSN PERFORM MASK-NAME PERFORM MASK-ACCOUNT-NUMBER PERFORM ADJUST-BALANCE WRITE MASKED-RECORD. MASK-SSN. * Replace all but last 4 digits with X's STRING "XXXXX" DELIMITED BY SIZE PR-SSN(6:4) DELIMITED BY SIZE INTO MR-SSN. MASK-NAME. * Replace real names with generic names EVALUATE PR-CUSTOMER-ID(10:1) WHEN "0" MOVE "TEST USER ZERO " TO MR-CUSTOMER-NAME WHEN "1" MOVE "TEST USER ONE " TO MR-CUSTOMER-NAME WHEN "2" MOVE "TEST USER TWO " TO MR-CUSTOMER-NAME WHEN "3" MOVE "TEST USER THREE " TO MR-CUSTOMER-NAME WHEN "4" MOVE "TEST USER FOUR " TO MR-CUSTOMER-NAME WHEN "5" MOVE "TEST USER FIVE " TO MR-CUSTOMER-NAME WHEN "6" MOVE "TEST USER SIX " TO MR-CUSTOMER-NAME WHEN "7" MOVE "TEST USER SEVEN " TO MR-CUSTOMER-NAME WHEN "8" MOVE "TEST USER EIGHT " TO MR-CUSTOMER-NAME WHEN "9" MOVE "TEST USER NINE " TO MR-CUSTOMER-NAME END-EVALUATE. MASK-ACCOUNT-NUMBER. * Replace with fixed test account numbers + last 4 of original STRING "TESTACCT" DELIMITED BY SIZE PR-ACCOUNT-NUM(9:8) DELIMITED BY SIZE INTO MR-ACCOUNT-NUM. ADJUST-BALANCE. * Round balances to nearest 100 to help mask exact values COMPUTE MR-BALANCE = (PR-BALANCE / 100) * 100.
Ensure your test data covers these critical areas:
| Test Criteria | Examples for COBOL Programs |
|---|---|
| Boundary Values | Maximum/minimum field values, zero values, one below/above limits |
| Data Types | Numeric, alphanumeric, packed decimal, binary, floating-point |
| Special Cases | Empty fields, non-printable characters, foreign languages, special symbols |
| Business Scenarios | Common transactions, historical cases, seasonal processing |
| Error Conditions | Invalid dates, out-of-range values, incorrect codes |
| Volume Testing | Empty files, single record, few records, maximum expected volume |
Regression testing ensures that modifications to a program don't break existing functionality. This is particularly important for COBOL systems that have evolved over decades and contain critical business logic.
| Approach | When to Use | Implementation in COBOL Environment |
|---|---|---|
| Full Regression | Major releases, critical systems | Run all test cases with complete input data sets |
| Partial Regression | Minor changes, time constraints | Test modified modules and those impacted by changes |
| Risk-Based Regression | Mixed criticality components | Focus on high-risk areas, critical business functions |
| Smoke Testing | Rapid verification after builds | Run critical path tests to verify basic functionality |
| Delta Testing | Specific feature changes | Test only what's changed using change impact analysis |
Automated regression testing for COBOL can be implemented in several ways:
1234567891011121314151617181920212223242526272829303132333435363738394041424344454647//REGTEST JOB (ACCT),'REGRESSION TEST',CLASS=A,MSGCLASS=X //* //* AUTOMATED REGRESSION TEST FOR CUSTOMER BILLING SYSTEM //* //STEP01 EXEC PGM=CBILLMN //STEPLIB DD DSN=TEST.LOADLIB,DISP=SHR //CUSTMAST DD DSN=TEST.REGRESSION.CUSTMAST,DISP=SHR //BILLHIST DD DSN=TEST.REGRESSION.BILLHIST,DISP=SHR //NEWTRANS DD DSN=TEST.REGRESSION.NEWTRANS,DISP=SHR //BILLOUT DD DSN=TEST.REGRESSION.ACTUAL.BILLOUT, // DISP=(NEW,CATLG,DELETE), // SPACE=(CYL,(1,1)), // DCB=(RECFM=FB,LRECL=100,BLKSIZE=0) //ERRLOG DD DSN=TEST.REGRESSION.ACTUAL.ERRLOG, // DISP=(NEW,CATLG,DELETE), // SPACE=(CYL,(1,1)), // DCB=(RECFM=FB,LRECL=80,BLKSIZE=0) //SYSOUT DD SYSOUT=* //* //STEP02 EXEC PGM=IDCAMS //SYSPRINT DD SYSOUT=* //SYSIN DD * COMPARE - DATA(TEST.REGRESSION.ACTUAL.BILLOUT) - WITH(TEST.REGRESSION.BASELINE.BILLOUT) - EXITCK(10) IF LASTCC > 0 THEN - DO SET MAXCC = 8 WRITE 'REGRESSION TEST FAILED: BILLOUT FILE MISMATCH' END COMPARE - DATA(TEST.REGRESSION.ACTUAL.ERRLOG) - WITH(TEST.REGRESSION.BASELINE.ERRLOG) - EXITCK(10) IF LASTCC > 0 THEN - DO SET MAXCC = 8 WRITE 'REGRESSION TEST FAILED: ERRLOG FILE MISMATCH' END /* //* //STEP03 EXEC PGM=TESTRPT,COND=(8,LT) //STEPLIB DD DSN=TEST.LOADLIB,DISP=SHR //DIFFILE DD DSN=TEST.REGRESSION.DIFFS,DISP=SHR //RESULTS DD SYSOUT=* //*
This JCL example runs a program with regression test data, compares outputs to baselines, and generates a report of differences if any are found.
Maintaining regression test baselines requires careful management:
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135* Example of a COBOL program to selectively compare files IDENTIFICATION DIVISION. PROGRAM-ID. SELCOMP. ENVIRONMENT DIVISION. INPUT-OUTPUT SECTION. FILE-CONTROL. SELECT BASELINE-FILE ASSIGN TO BASELIN ORGANIZATION IS SEQUENTIAL. SELECT ACTUAL-FILE ASSIGN TO ACTUALIN ORGANIZATION IS SEQUENTIAL. SELECT DIFF-FILE ASSIGN TO DIFFOUT ORGANIZATION IS SEQUENTIAL. DATA DIVISION. FILE SECTION. FD BASELINE-FILE RECORDING MODE IS F RECORD CONTAINS 100 CHARACTERS. 01 BASELINE-RECORD. 05 BR-CUSTOMER-ID PIC X(10). 05 BR-TRANSACTION-TYPE PIC X. 05 BR-AMOUNT PIC 9(9)V99. 05 BR-DATE PIC X(8). 05 BR-TIME PIC X(8). 05 BR-RESULT-CODE PIC X(2). 05 BR-RESULT-DESC PIC X(50). 05 BR-FILLER PIC X(10). FD ACTUAL-FILE RECORDING MODE IS F RECORD CONTAINS 100 CHARACTERS. 01 ACTUAL-RECORD. 05 AR-CUSTOMER-ID PIC X(10). 05 AR-TRANSACTION-TYPE PIC X. 05 AR-AMOUNT PIC 9(9)V99. 05 AR-DATE PIC X(8). 05 AR-TIME PIC X(8). 05 AR-RESULT-CODE PIC X(2). 05 AR-RESULT-DESC PIC X(50). 05 AR-FILLER PIC X(10). FD DIFF-FILE RECORDING MODE IS F RECORD CONTAINS 200 CHARACTERS. 01 DIFF-RECORD PIC X(200). WORKING-STORAGE SECTION. 01 FILE-STATUS. 05 BASELINE-EOF PIC X VALUE 'N'. 88 END-OF-BASELINE VALUE 'Y'. 05 ACTUAL-EOF PIC X VALUE 'N'. 88 END-OF-ACTUAL VALUE 'Y'. 01 COUNTERS. 05 RECORDS-COMPARED PIC 9(7) VALUE ZERO. 05 RECORDS-DIFFERENT PIC 9(7) VALUE ZERO. 01 COMPARISON-RESULT PIC X. 88 RECORDS-MATCH VALUE 'Y'. 88 RECORDS-DIFFER VALUE 'N'. 01 DIFF-LINE. 05 DL-HEADER PIC X(30). 05 DL-BASELINE PIC X(80). 05 FILLER PIC X(10) VALUE SPACES. 05 DL-ACTUAL PIC X(80). PROCEDURE DIVISION. MAIN-PARAGRAPH. OPEN INPUT BASELINE-FILE OPEN INPUT ACTUAL-FILE OPEN OUTPUT DIFF-FILE PERFORM UNTIL END-OF-BASELINE OR END-OF-ACTUAL READ BASELINE-FILE AT END SET END-OF-BASELINE TO TRUE END-READ IF NOT END-OF-BASELINE READ ACTUAL-FILE AT END SET END-OF-ACTUAL TO TRUE END-READ IF NOT END-OF-ACTUAL ADD 1 TO RECORDS-COMPARED PERFORM COMPARE-RECORDS IF NOT RECORDS-MATCH ADD 1 TO RECORDS-DIFFERENT PERFORM WRITE-DIFF-RECORD END-IF END-IF END-IF END-PERFORM CLOSE BASELINE-FILE CLOSE ACTUAL-FILE CLOSE DIFF-FILE DISPLAY "Records compared: " RECORDS-COMPARED DISPLAY "Differences found: " RECORDS-DIFFERENT STOP RUN. COMPARE-RECORDS. * Compare records, ignoring date and time fields SET RECORDS-MATCH TO TRUE IF BR-CUSTOMER-ID NOT = AR-CUSTOMER-ID OR BR-TRANSACTION-TYPE NOT = AR-TRANSACTION-TYPE OR BR-AMOUNT NOT = AR-AMOUNT OR BR-RESULT-CODE NOT = AR-RESULT-CODE OR BR-RESULT-DESC NOT = AR-RESULT-DESC SET RECORDS-DIFFER TO TRUE END-IF. WRITE-DIFF-RECORD. MOVE "RECORD #:" TO DL-HEADER MOVE BASELINE-RECORD TO DL-BASELINE MOVE ACTUAL-RECORD TO DL-ACTUAL MOVE DIFF-LINE TO DIFF-RECORD WRITE DIFF-RECORD IF BR-CUSTOMER-ID NOT = AR-CUSTOMER-ID MOVE "CUSTOMER ID DIFFERS:" TO DL-HEADER MOVE BR-CUSTOMER-ID TO DL-BASELINE MOVE AR-CUSTOMER-ID TO DL-ACTUAL MOVE DIFF-LINE TO DIFF-RECORD WRITE DIFF-RECORD END-IF * Similar comparisons for other fields .
For large systems, identifying which components need regression testing is crucial: