Starting from:

$30

CS1632-Exercise 1 Solved

For this exercise, you will determine a test plan for the
simple simulator GoatGoatCar, based on the requirements listed.  There should
be at least 6 test cases altogether.  Test cases should mention all necessary
preconditions, execution steps, and postconditions.  Additionally, a
traceability matrix showing the mapping of test cases to requirements is
required.  

There are several known defects in the software (10 at last count); try to find and
report on at least two.

It is expected that you execute the test plan in order to find the defects, but
you may find it useful to initially do some exploratory testing to determine
how the system works.  While you are not expected to find *all* of the defects,
a reasonable test plan should definitely find at least two.  This is an
intentionally target-rich environment.

You may ask me any questions you may have during class.  It's not scored.  Ask me anything.

## Creating a Test Plan

Remember, the template for test cases -

```
     IDENTIFIER: [A unique number or string (e.g. TEST-ARGS-NUMBER-FIVE-ARGS)]
     TEST CASE: [A description of the test case]
     PRECONDITIONS: [State of the system before performing execution steps]
     EXECUTION STEPS: [Steps to perform test]
     POSTCONDITIONS: [ *EXPECTED* state of the system after performing execution steps]
```

The IDENTIFIER is some value which will uniquely specify the test case.  We
learned it can be either a number, or a more descriptive label (e.g.
TEST-INVALID-TIMES, TEST-LOW-NUM-TIMES, etc.).  For this exercise, please use a
descriptive label.  Note that the INPUT VALUES and OUTPUT VALUES fields in the
template are omitted because we are not doing method unit testing.
Please refer to [Lecture 4: Test Plans](../../lectures/CS1632_Lecture4_Test_Plans.pdf) Slides 8 - 13
for more details and examples for each item.

PRECONDITIONS is the state of the system before performing the test.  If the
system is a website, it is things like: "user is logged into the website" or
"user is subscribed to the mailing list", etc.  The program we will be testing
today is a commandline Java program where the test is simply running the
program with a set of arguments.  There is no program state to speak of before
running the test.  Then, what things should you put as preconditions?  There
are plenty!  For a Java program, it is always important which Java Runtime
Environment (JRE) the program runs in.  You could say something like: When
"java -version" is run, the system outputs "java version "1.8.0_231".
Companies would actually perform multiple tests using the same arguments for
multiple versions of Java that their clients use, and each would be a separate
test case!

EXECUTION STEPS should be numbered step-by-step instructions on what you expect
the tester to do.  It should be exact to the letter so that tests are
reproducible.  The tested program does not allow any interaction with the user
other than to run it with commandline arguments, so all steps will be
one-liners in this case, but typically there will be multiple steps.

POSTCONDITIONS should be the **expected** state of the program after the test.
It is **not** the observed state.  So where should you get expectations of your
program?  From the [requirements](requirements.md) of course!

## Creating a Traceability Matrix

A traceability matrix allows us to correlate test cases to requirements and
vice versa.  It allows us to check why a test case is being run (to check one
or more requirements).  It also allows us to check how much testing a
particular requirement is receiving.  See Chapter 6 section 6 (6.6) in the
textbook for examples and details on creating them.

Note that some test cases may test several requirements at once.  This is only
natural.  As we saw on the last example on the lecture slides, in a real-world
traceability matrix, you can have one test case mapped to multiple
requirements.  Vice versa, you can have several test cases for one requirement.

Also note that a good traceability matrix must cover all requirements to have
no gaps in test coverage.  Please make sure of this.

## Reporting Defects

Please listen to "Lecture 5: Defects" before completing this section.

This is the correct format for defects -

```
      IDENTIFIER: [A unique number or string (e.g. BUG-ARGS-NUMBER-FIVE-ARGS)]
      SUMMARY: [A one sentence description of defect]
      DESCRIPTION: [A detailed description of everything the tester discovered]
      REPRODUCTION STEPS: [Preconditions + Steps to reproduce (similar to test case execution steps)]
      EXPECTED BEHAVIOR: [What you expected according to requirements]
      OBSERVED BEHAVIOR: [What you *ACTUALLY* saw]
```

Please refer to [Lecture 5: Defects](../../lectures/CS1632_Lecture5_Defects.pdf) Slides 15 - 22
for more details and examples for each item.  Optional bug report items
such as SEVERITY or IMPACT were not included for this exercise.

**Don't forget to include any preconditions at the start of the REPRODUCTION
STEPS.** You will not be able to reproduce the bug even if you reproduce the
steps if you start from a different precondition!

## Test Application: GoatGoatCar

GoatGoatCar is going to be our way of determining the correct answer to the
"Monty Hall Problem" (https://en.wikipedia.org/wiki/Monty_Hall_problem).  The
Monty Hall Problem can be summarized in pictures:

<img alt="Figure 1" src=img/monty-hall-pic-1.jpg width=350
<img alt="Figure 2" src=img/monty-hall-pic-2.jpg width=350
<img alt="Figure 3" src=img/monty-hall-pic-3.jpg width=350
<img alt="Figure 4" src=img/monty-hall-pic-4.jpg width=350
<img alt="Figure 5" src=img/monty-hall-pic-5.jpg width=350

What do you think?  The answer lies in the below illustration:

<img alt="Figure 6" src=img/monty-hall-pic-6.svg width=350

If you still don't get it, you can read the above wikipedia entry if you like
reading, or here is a video if you are a more visual learner: https://www.youtube.com/watch?v=9vRUxbzJZ9Y.

Another way to get at the answer is through Monte Carlo simulations, that is
simulate the game thousands of times to empirically answer the question one
way or another.  The program we will test today randomly generates games,
plays them one way another, and gives a summary at the end of what percentage
of the time switching would give you the "good item" and what percentage of
the time staying would have won you the "bad" item.

The program will accept four arguments, in this order:

* __"Good" option__ - This will be the item that the player actually wants (e.g., a car).

* __"Bad" option__ - This will be the item that the player does not want (e.g., a goat)

* __Num Times__ - This is the number of rounds that will be simulated.

* __Num Threads__ - This will be the number of threads that the simulation runs in

Example command to execute.  This will use "car" as the good choice, "goat" as
the bad choice, 10001 as the number of rounds to simulate, and it will do it on
four threads.

```
$ java -jar GoatGoatCar.jar car goat 10001 4
```

GoatGoatCar.jar is available in this directory.  

Create a reasonable test plan based on the [requirements](requirements.md).
Hint: Try to have a combination of explicit boundary values and implicit
boundary values as well as interior values in your test cases.  As we learned,
this is where most of the defects will reside!

More products