Jacquard Example 1

Jacquard [Javadoc] is a Java autograder library for Gradescope. This repository provides an example of a Jacquard-based autograder that was used for a quiz in a Java data structures course. It demonstrates these features, applied to a single submitted file.

• Checkstyle tests
• PMD tests
• Parser-based tests (ensuring that language features are used as required)
• JUnit 5 tests

The only significant change from Jacquard Example 0 is the addition of parser-based tests, which use syntax-based graders.

Prerequisites

The rest of this document assumes that you have read:

• Jacquard Example 0, a more detailed introduction to the basic features
• The student instructions for this assignment

Video Explanation

1. A simple syntax-based grader (2:56)
2. A complex syntax-based grader

Note

The remaining content is the same as for Jacquard Example 0, except for changes to file names.

Software Requirements

• bash (included on OS X and Linux) for testing and building the autograder
• Python 3 if you want to test locally by executing test_autograder.py (optional)

For bash on Windows, we use and recommend Git for Windows 2.41 or higher.

Files

These directories and files have code specific to the assignment:

• config/checkstyle-rules.xml holds the checkstyle rules file
• src/main/java/student contains
  • AutograderMain.java, which has the main method that controls the autograder
  • FavoritesIterator.java, placeholder for student code for testing the autograder locally
  • HiddenFavoriteIteratorsTest.java and ProvidedFavoritesIteratorTest, which contain JUnit 5 tests of student code
• submission holds a sample submission (required if you want to run test_autograder.py locally)
• submissions (which is not required) holds sample submissions to manually test the grader locally or on Gradescope
  • perfect, a subdirectory with a perfect submission
  • imperfect, a subdirectory with an imperfect submission
  • starter, a subdirectory with the starter code

Any of the above files could have different names or packages, although package names must currently have only a single part (e.g., student, not edu.myschool.student.) The submissions/ subdirectories could also have different names.

config.ini

The submission package and files are specified in config.ini:

[submission]
package = student
files = [FavoritesIterator.java]

The list of files is comma-separated, with optional whitespace.

build.gradle

The main class of the autograder is specified in build.gradle:

application {
    mainClass.set("student.AutograderMain")
}

You will need to change it if you use a different package/class name for your main autograder class.

You are free to make other additions to build.gradle, such as adding dependencies.

Dockerfile

This can be used to create a Docker image for the autograder.

Shell scripts

test_autograder.sh

This script, which requires Python 3, lets you test the autograder locally. If called without any arguments, it will use the submission in the submission/ directory.

./test_autograder.sh

If called with an argument, it will use the submission in that subdirectory, prepending submissions/ if necessary. For example, to test the autograder against the files in submissions/perfect, you could execute either:

./test_autograder.sh submissions/perfect
./test_autograder.sh perfect

make_autograder.sh

This creates the zip file for you to upload to Gradescope.

Uploading to Gradescope

You can provide the grader to Gradescope as a zip file (recommended) or as a Docker image.

Zip file

To create a zip file, run ./make_autograder.sh from the command line.

To configure the autograder on Gradescope:

1. Click on "Configure Autograder" in the left sidebar.
2. Select "Zip file upload".
3. Click on "Replace Autograder (.zip)".
4. Select:
   • Base Image OS: Ubuntu
   • Base Image Version: 22.04
   • Base Image Variant: JDK 17
5. Click on "Update Autograder". (You may have to wait up to a minute for anything to happen. The button will go gray when the build begins.)
6. Wait for the "Built as of" time to be updated.

Docker

If you have installed Docker, you can create an image based on Dockerfile with a command line of the form:

# docker build -t username/gradername .

For example, because my Docker username is espertus, I would type:

docker build -t espertus/example1 .

For more information, see Manual Docker Configuration.

Next Steps

See https://github.com/jacquard-autograder/jacquard for more information, including:

• Documentation
• Additional examples
• Mailing lists