Platform System Design

This project is a part of a bigger picture. While coder FE (front-end) serves the users and allows them to upload codes to the platform, coder BE(back-end) will push the codes alongside with the problem info into a queue. Then, Judge will independently read from this queue, schedule jobs to compile codes and then run them against test cases.

I was inspired by Kubernetes in this design. Services could independently write their reached/desired state into a shared file system and read from that file system for their next jobs.

Coder Judge

Coder Judge is an open-source platfrom written in Java and Python to compile and run code submissions against test-cases. It uses RabbitMQ to fetch pending jobs, compiles them and then runs agains test cases. After judging, one of these verdics are possible for the code.

• COMPILE-ERROR
• ACCEPT
• WRONG-ANSWER
• RUNTIME-ERROR
• TIME-LIMIT
• MEMORY-LIMIT

Currently, we support C++ and Python as the submission languages. You can add other languages by simply adding compiler in app/other/compile.py and modifying the runner script in app/other/run.py.

Judge Setup

You need java and gradle on your system to run this project.

Environment file pattern

The .env file must be placed in the root directory following the pattern described below.

DB_URL=                    # database host address (including port)
DB_USER=                   # database username
DB_PASSWD=                 # database password

QUEUE_HOST=                # queue host address (including port)
QUEUE_USER=                # queue username 
QUEUE_PASSWD=              # queue password
SUBMISSION_QUEUE_NAME=     # queue name for submissions
RESULT_QUEUE_NAME=         # queue name for results

FILE_PATH=                 # root filesystem path

Setup envirounment variables

You need to set-up every variable on your .env file to your current envirounment. This can simply done with Docker, or on bash with:

$ export $(grep -v '^#' .env | xargs)

Run the project

./gradlew run