jpa01 : Testing: test coverage and mutation testing

num ready? description assigned MW lect due MW lect assigned TR lect due TR lect
jpa01 true Testing: test coverage and mutation testing Tue 09/27 01:00PM Sun 10/09 11:59PM

This assignment is jpa01, i.e “Java Programming Assignment 01”.

If you find typos or problems with the lab instructions, please report them on the #typos channel on Slack

In this lab:


Before starting this lab, whether you are brand new to Java, or even if you are very experienced with Java, we encourage you to read through parts 1 through 12 of this Java tutorial:

Most of this lab is concerned with really basic object-oriented programming: getters, setters, and simple methods of a class.

However, there are a few concepts that may be new even to the most experienced Java programmers:

It covers some concepts that may be new to you, even if you have worked with Java before, including:

You may also want to read ahead in the Head First Java book (HFJ2), or use the Java in a Nutshell v7 text (JN7) if you encounter Java concepts with which you are unfamiliar.

Working in a pair?

You have the option to work in a pair on this assignment.

To set up a pair repo, see these instructions: Pair Repo Setup.

In addition:

If you are in your repo directory, and type git log at the command line, you’ll see a list of the commits for your repo.

Record that you are pairing on each commit message by putting the initials of the pair partners at the start of the commit message.

E.g. If Selena Gomez is driving, and Justin Timberlake is navigating, and you fixed a bug in your getDanceMoves() method, your commit message should be SG/JT fixed bug in getDanceMoves()

We should see frequent switches between SG/JT and JT/SG.

When remote pairing, this means switching who is sharing their screen:


Step 0: Set up your repo

When cloning your repo, be sure to use only the ssh links.

You may work individually or as a pair on this lab. However, if you work as a pair, please:

The starter code is in Visit that page for the approrpiate URL to add the starter remote.

To add the starter as a remote, cd into the repo you cloned, then do:

git remote add starter

Then do:

git checkout -b main
git pull starter main
git push origin main

That should get you set up with the starter code.

Step 1: Get oriented to packages

A few things to notice:

Don’t change the package from pconrad to your name; the Gradescope autograder is looking for the code under the edu.ucsb.cs56.pconrad.menuitems package. So each source file:

Here are the commands you’ll need as you work with the code. Try them out now.

To do this Type this command Notes
compile the code mvn compile  
reset everything mvn clean  
run the tests mvn test  
generate a report of test coverage mvn test jacoco:report Open the file target/site/jacoco/index.html in a browser to read the report. (Note that a report will only be generated if your unit tests pass)
generate a jar file mvn package  
generate a mutation test report mvn test org.pitest:pitest-maven:mutationCoverage Open the file target/pit-reports/yyyymmddhhmm/index.html to see the report; note that yyyymmddhhmm will be replaced by a date/timestamp.

Step 2: Start writing code to make tests pass

In this lab, you’ll be implementing several methods of a class called MenuItem that represents item on a restaurant Menu.

(There is a follow up lab in which we will add a Menu class that uses these menu items; but we need to discuss sorting, java.lang.Comparable, java.util.Comparator, and Java lambda expressions in lecture first before we get to that.)

A MenuItem represents an item on the menu of a restaurant. It has three attributes:

Note that the starter code:


So you’ll need to do a bit more work than you may be used to.

I suggest that you work in this order:

Let me say this again, because I’ve assigned this lab a few times now, and each time, students fail to read this important instruction:

YOU MUST WRITE STUBS FOR ALL METHODS before any of the Gradescope tests will work. Until you write stubs for all methods, the code will FAIL TO COMPILE on Gradescope, and no tests will pass.

Details about methods of MenuItem

The constructor has the signature:

public MenuItem(String name,
                int priceInCents,
                String category)

Here are the instance methods you’ll need to implement for MenuItem

Modifier and Type Method Description
String getCategory() Returns the category of the menu item
String getName() Returns the name of the menu item
String getPrice() Returns the price, formatted as a string with a $.
String getPrice(int width) Returns the price, formatted as a string with a $, right justified in a field with the specified width.
int getPriceInCents() get the price in cents only
String toString() return a string in csv format, in the order name,price,cateogry.
For example: "Small Poke Bowl,1049,Poke Bowls"
In this case, the price is unformatted; just an integer number of cents.

Checking the tests

To check the code against the tests that you’ve written, use:

Interpreting the output can be

Step 3: Learning about Test Coverage

Now, did you really write unit tests for all of your code? Let’s check!

We can automatically compute “test case coverage”, using a tool call JaCoCo (Java Code Coverage).

Read these short articles about test coverage before moving to step 4:

Once you’ve looked over those, it’s time to check your test coverage, which we’ll do in Step 4.

Step 4: Checking Test Case Coverage

Be sure that you’ve added your pair partner to your submissions on Gauchospace

Then, check your test coverage!

Checking Test Case Coverage Locally.

How you go about that depends a lot on your setup.

How to read the test coverage reports

Step 5: Try to get as close to 100% coverage as you can

Keep reworking your code until you get as close as you can to 100% test coverage.

As we’ll discuss in lecture, 100% coverage isn’t always necessary or even desirable, but in this case you should be able to get there, or at least pretty close.

Try to get as close to 100% test coverage as you can. Then, once you’ve gotten to 100%, we’ll check the quality of your tests with an even more rigorous method, called “mutation testing”.

What is mutation testing?

Here is an explanation of mutation testing from the website for pitest which is the package we are using for mutation testing.

Explanation of Mutation Testing (click the arrow to reveal)

What is mutation testing?

How it works in 51 words

Mutation testing is conceptually quite simple.

Faults (or mutations) are automatically seeded into your code, then your tests are run. If your tests fail then the mutation is killed, if your tests pass then the mutation lived.

The quality of your tests can be gauged from the percentage of mutations killed.


Really it is quite simple

To put it another way - PIT runs your unit tests against automatically modified versions of your application code. When the application code changes, it should produce different results and cause the unit tests to fail. If a unit test does not fail in this situation, it may indicate an issue with the test suite.


What’s wrong with line coverage?

Traditional test coverage (i.e line, statement, branch, etc.) measures only which code is executed by your tests. It does not check that your tests are actually able to detect faults in the executed code. It is therefore only able to identify code that is definitely not tested.

The most extreme examples of the problem are tests with no assertions. Fortunately these are uncommon in most code bases. Much more common is code that is only partially tested by its suite. A suite that only partially tests code can still execute all its branches (examples).

As it is actually able to detect whether each statement is meaningfully tested, mutation testing is the gold standard against which all other types of coverage are measured.

So, to clarify, you will run mutation testing once you’ve gotten to 100% line/branch coverage, or as close to it as you can get. At that point, you run:

mvn test org.pitest:pitest-maven:mutationCoverage

(Yeah, no one is going to expect you to memorize that command. We might expect you to know mvn compile, mvn package, and mvn test, and maybe even mvn test jacoco:report, but not this one.)

When you run that command, your file will be mutated; the pitest software will create mutant versions of that code in an attempt to deliberately introduce bugs. A mutant is killed when your test suite finds the bug. It survives when your test suite doesn’t kill the bug.

The output will show you how many mutants are killed.

You can see more detailed output by using a web browser to open up the file:

Note that yyyymmddhhmm will be replaced by a date/timestamp; each time you run the Maven command to run pitest, it will produce a new version of the report with a new timestamp.

It may be convenient to use mvn clean before running a pitest mutation report; that way there will only be one such directory rather than multiple ones.

Another way to see the Pitest report: GitHub Actions artifacts

Another way to see the pitest report is to let the GitHub Actions run for your repo (which happens each time you push a change to GitHub). When you do, you should see that there is a link for “Artifacts” when you examine the Github Actions results.

Note: If you don’t see the Artifacts as shown below, click the arrow for hints on what might be wrong.

First, you must be passing all of the regular JUnit tests, or you won’t get a mutation testing report from pitest.
So check that you are passing all of the regular JUnit tests first.

Second, there is a chance that you picked up an old version of the starter code—and to be clear, you didn’t necessarily do anything wrong; the first version of the instructions had master instead of main, and that branch didn’t have everything in it.

To check, compare your .github/workflows/maven.yml file with the listing of the file here: .github/workflows/maven.yml from starter code

Specifically, make sure the files ends with these lines, indented so that all of the - name commands line up (just like in the starter code at the link above)”:

      - name: Pitest
        run: mvn test org.pitest:pitest-maven:mutationCoverage
      - name: Upload Pitest to Artifacts
        uses: actions/upload-artifact@v2
          name: pitest-mutation-testing
          path: target/pit-reports/**/*  

If you download the artifacts, you’ll get a .zip file that you can download, and open. Inside that zip file, you’ll find an index.html file that you can open in a web browser. That will show you the Pitest mutation report.

click the green check

click where it says details

click on artifacts

click to download zip

Mutation testing is 30% of your grade

30% of your grade for this lab will come from the number of mutants you kill. Kill 100% of the mutants, you get 30/30. Kill only half of them, you get 15/30.

Gradescope will run the mutation testing on your solution, and award points proportionately depending on how many you kill.

Now: since the mutants are based on your your individual solution, it is impossible to know in advance how many there will be, or what they will be. But you can investigate by looking at the report.

Step 6: Before final submission on Gradescope

Look for jpa01 on Gradescope.

If working in a pair:

End of description for jpa01