|
|
|
John D. McGregor |
|
Dept of Computer Science |
|
Clemson University |
|
|
|
|
|
Unit |
|
A fundamental unit, such as a class, is tested
to determine whether it fulfills its post-conditions |
|
|
|
Subsystem |
|
A set of integrated units are tested to be
certain that they play nicely together |
|
|
|
System |
|
A completed application is tested as it will be
used |
|
|
|
|
|
Functional |
|
Tests are based on the specification of WHAT the
item does |
|
|
|
Structural |
|
Tests are based on the structure that defines
HOW the unit operates |
|
|
|
Interaction |
|
Tests are based on the communication that
happens between units |
|
|
|
|
|
Input-based |
|
Analyze the data types of parameters and
variables based on the problem domain |
|
Partition into “equivalence classes” |
|
Select inputs from each equivalence class |
|
setTemperature(int degrees) if the domain refers
to water, the equivalence classes might be degrees<32;
32<=degrees<=212;degrees>212 |
|
We assume that any value between 32 and 212 will
cause the program to use the same mechanism so if one value is computed
correctly then all will be |
|
|
|
|
|
|
|
Output-based |
|
Based on possible outcomes |
|
Win game |
|
Lose game |
|
Select test data to achieve these outcomes |
|
|
|
|
void doIt(int x) |
|
|
|
If you select one value for x, execute doIt for
that value, and get a correct answer, how confident are you that doIt
works? |
|
|
|
What if you run 50 values through and get
correct answers? Feel better? |
|
|
|
The more completely a domain is covered by test
cases, the more likely the result is to be correct. |
|
|
|
|
Testing operates in parallel to the development
process |
|
|
|
Unit tests are written and applied by the
developer |
|
They are applied when the developer is ready |
|
|
|
Subsystem tests are often written and applied by
an “integration” team |
|
These integrations or “builds” are scheduled
when all the classes in a subsystem are ready |
|
|
|
System tests are written and applied by a system
test team. |
|
Scheduled when sufficient end-user functionality
is available to test |
|
|
|
|
Based on the public interface of the unit |
|
|
|
public int largest(int a, int b, int c) |
|
|
|
Test cases sample from the “domain” of the input |
|
Possible inputs are divided into equivalence
classes |
|
|
|
a b c correct answer |
|
1 3 5 5 |
|
1 5 3 5 |
|
5 1 3 5 |
|
5 3 1 5 |
|
|
|
|
>javac –classpath .;%classpath% trial5.java |
|
|
|
>java –classpath .;%classpath% trial5 |
|
|
|
5 |
|
|
|
Repeat this using all the tests on the previous
page |
|
|
|
Are we confident this method is correct? |
|
|
|
|
public class trial5{ |
|
public trial5(){ |
|
} |
|
public int largest(int a,int b,int c){ |
|
return 5; |
|
} |
|
|
|
public static void main(String[] args){ |
|
trial5 t = new trial5(); |
|
System.out.println(t.largest(1,3,5)); |
|
} |
|
} |
|
|
|
|
|
|
Test cases must establish the pre-conditions |
|
Test cases are developed using the specified
outcomes in the post-condition |
|
|
|
Puck PuckSupply::getPuck() |
|
pre:
true |
|
post:
result = pucks->asSequence->first and size = size@pre - 1 |
|
|
|
|
|
|
Tests are sequenced to “cover” the transitions |
|
|
|
|
Based on the structure of the code |
|
|
|
Public int largest(int a, int b, int c) |
|
{ |
|
if (a > b){ |
|
if(a > c){ |
|
return a; |
|
} |
|
else{ |
|
if(c > b){ |
|
return c; |
|
} |
|
} |
|
else return 0; |
|
} |
|
|
|
|
|
|
In object-oriented software, integrated units
may still be a single object: |
|
|
|
|
|
|
|
|
Predominantly functional |
|
|
|
Based on use cases |
|
|
|
Additional test cases are targeted at specific
non-functional requirements such as performance, security, … |
|
|
|
|
|
|
|
|
|
|
|
Inputs: mouse – where can it be moved |
|
|
|
Test cases: |
|
Input-based |
|
Paddle hits puck on first descent |
|
Paddle allows puck to hit floor |
|
Output-based |
|
Paddle
hits puck several times in a row but then loses it and eventually
loses all pucks – loss |
|
Paddle hits puck many times – eventually game is
won |
|
|
|
|
Just like physical objects, software wears
overtime |
|
Changes in DLLs and operating systems lead to
this wear |
|
This leads to the need for self-tests |
|
|
|
|
|
|
Performance tests |
|
|
|
Security tests |
|
|
|
Certification tests |
|
|
|
Acceptance tests |
|
|
|
Deployment tests |
|
|
|
|
|
|
|
|
Suppose we have 3 variables |
|
And that each variable can take on 3 values |
|
|
|
27 possible combinations |
|
|
|
1,1,1 |
|
1,1,2 |
|
1,1,3 |
|
1,2,1 |
|
1,2,2 |
|
1,2,3 |
|
… |
|
|
|
|
|
Notes