Test Data & Finding Errors (Cambridge (CIE) A Level Computer Science): Revision Note
Exam code: 9618
Types of test data
What test data does Cambridge expect?
The 9618 syllabus names three types of test data: normal, abnormal and extreme/boundary
Cambridge treats extreme and boundary as one category, and mark schemes accept either word
If you have studied GCSE Computer Science, note that the word "erroneous" does not appear in 9618 mark schemes; use abnormal instead
Type of test data | What it is | Expected result | Example for a range of 150 to 155 inclusive |
|---|---|---|---|
Normal | Data inside the acceptable range, of the correct type | Accepted and processed | 153 |
Extreme/boundary | The largest and smallest acceptable values | Accepted | 150 and 155 |
Abnormal | Data that should be rejected, such as a value outside the range | Rejected | 149, 156 or 200 |
Values just outside the range, such as 149 and 156, test the boundary from the other side
Mark schemes accept these as abnormal or boundary, as long as the expected result is "rejected"
The marks come from the value and its expected result; the label alone rarely scores
How to choose test data for a validation rule
How do you write a test plan in the exam?
Most 9618 test plan questions give one completed row and ask you to "complete the table by writing additional tests", usually for 4 marks, one per row
Follow these steps:
Find the exact limits: "24 to 37, inclusive" means 24 and 37 are both valid; examiners note that "30 to 40" also includes both boundary values
Test each boundary from both sides: the lowest valid value, the highest valid value and the value just outside each one
Give the expected result in the form the code produces, such as
TRUE,FALSE,"PASS"or"FAIL"Make every test different: each row must test something new
Keep every test value the data type the question states; if the function takes an integer, real numbers and characters are not valid test data for that question
Some questions ask for test data sequences for a loop that ends on a terminator value, such as 99; each sequence should end with the terminator and test something different, such as only odd values, only even values, or just the terminator with no values before it
What mistakes do examiners see in test plans?
Examiner reports for 9618 Paper 2 highlight these mistakes:
Repeating normal data: for a range of 150 to 155, students gave several values from 151 to 154, which all test the same thing
Ignoring the stated data type: when the question said the sensor produces an integer, many students gave characters and real numbers
Copying the example: using test data that was too similar to the example given in the question was a common mistake
Types of error
What are syntax, logic and run-time errors?
Syntax error: a statement breaks the rules (grammar) of the language, so the program cannot be translated
Example:
NEXT Indexused to end aWHILEloop, instead ofENDWHILE
Logic error: the program runs but does not produce the expected result
Example:
IF Num > Min THENin an algorithm that should find the minimum
Run-time error: the program fails while it is running, for example by crashing or stopping responding
Example: accessing
Data[0]when the array is declaredARRAY[1:200], or dividing by zero
How do examiners ask about errors?
Questions give pseudocode that contains a stated number of errors, and ask you to find them, for example:
"The pseudocode contains two syntax errors and one other error. Identify the errors." [3]
"The pseudocode contains three errors due to the incorrect use of operators. Identify each error and state the correction required." [3]
"Write the three statements and explain how each could generate a run-time error."
When a question asks about errors in a given scenario, answer about that code; examiners report that some students gave only a textbook definition of a syntax error, without relating it to the scenario
Examiner Tips and Tricks
In one recent series, many students found all three errors in a piece of pseudocode but lost marks by putting them in the wrong category, so decide the category of each error before you write it
Identifying and correcting errors in given code
How do you find errors in pseudocode?
Check the code in this order:
Structure: every
IFhas anENDIF, everyWHILEhas anENDWHILE, everyFORhas aNEXTwith the same variableOperators:
←for assignment and=for comparison;&joins strings and+adds numbersArray indexes: every index stays within the declared bounds, including the first and last time round the loop
Conditions:
>vs>=,ANDvsOR, and whether the loop ends at the right pointInitial values: whether a starting value such as
Min ← 999could give a wrong answer for some inputs
Write your answer in the form the question asks for
If it asks you to identify the error, quote or describe the faulty statement
If it also asks for the correction, give the corrected statement as well
Examiners report that some students only gave a corrected line when the question asked them to identify the error
Example: errors due to operators
The pseudocode below should find the minimum of a series of numbers and count them
Min ← 999
Count ← 0
REPEAT
INPUT NextInput
IF IS_NUM(NextInput) = TRUE THEN
Num ← STR_TO_NUM(NextInput)
IF Num > Min THEN
Min ← Num
ENDIF
Count ← Count & 1
ENDIF
UNTIL NextInput ← "END"
The three operator errors, and their corrections, are:
Num > Minshould beNum < MinCount & 1should beCount + 1NextInput ← "END"should beNextInput = "END"
Worked Example
A function ValidTickets() will:
take an integer value as a parameter
return
TRUEif the value is within the range 1 to 8, inclusiveotherwise return
FALSE
(a) Complete the table to define a test plan to thoroughly test the function. The first row has been completed.
Type of test data | Test data value | Expected result |
|---|---|---|
Normal | 5 | TRUE |
[4]
(b) The function is written in pseudocode:
FUNCTION ValidTickets(Number : INTEGER) RETURNS BOOLEAN
IF Number > 1 AND Number <= 8 THEN
RETURN TRUE
ELSE
RETURN FALSE
ENDIF
ENDFUNCTION
The function contains a logic error. Identify which test from your test plan would reveal the error and state the correction required.
[2]
Answer:
(a) One mark per row:
Extreme/boundary, 1, TRUE [1 mark]
Extreme/boundary, 8, TRUE [1 mark]
Abnormal (or boundary), 0, FALSE [1 mark]
Abnormal (or boundary), 9, FALSE [1 mark]
(b) The test with value 1 would return FALSE instead of TRUE [1 mark]
Number > 1 should be Number >= 1 [1 mark]
Only the boundary test reveals the error, because 5, 8, 0 and 9 all give the expected result even with the faulty condition.
Testing methods
Which testing method should you name?
The 9618 syllabus lists nine testing methods; questions often describe a situation and ask you to name or describe the method
Dry run: working through the code by hand, often by completing a trace table
Walkthrough: going through the code or algorithm step by step, often with other people, to check its logic
White-box testing: testing every possible path through the code, using knowledge of the code itself
Black-box testing: testing that inputs give the expected outputs, without looking at the code
Stub testing: replacing modules that are not yet written with simple dummy modules that return a known result or output a message to show they have been called
Integration testing: testing modules together once each one has been shown to work on its own
Alpha testing: testing by the developer's own team before the software is released to users
Beta testing: testing by a small group of potential users, who report problems and suggest improvements
Acceptance testing: testing by the client to check that the finished program meets the agreed requirements
Stub testing and integration testing are often confused
Stub testing happens before a module exists, using a dummy in its place
Integration testing happens after modules are written and individually tested, by combining them
Unlock more, it's free!
Join the 100,000+ Students that ❤️ Save My Exams
the (exam) results speak for themselves:
Was this revision note helpful?
