University showcase15min
RIVETS: Under the Hood of a Game that Teaches (Clean) Coding — Language Pipeline, Level Format and Editor
RIVETS is a beginner-friendly programming game where players control a robot through 15 mazes using a tiny custom language. It rewards clean, abstract code like loops and nesting. The talk demos the game and editor, then explains its parser, AST, interpreter, scoring, level format, and AI-assisted development.
talk.summaryAiDisclaimer
Archontis E. KostisUniversity of Macedonia
talkDetail.whenAndWhere
Saturday, November 21, 14:45-15:00
Room 3 - Alexandros
Content:
RIVETS is a game that teaches programming to complete beginners: players write short programs in a tiny custom language to guide a robot through 15 maze levels. Reaching the goal is not enough: the game scores the player's code and rewards clean, abstracted solutions (loops and nesting) over brute-force sequences of moves, so players are nudged toward writing better code from level one. It was built with Java and Greenfoot as a Master's course assignment. The talk will include a live demo of the game and the level editor, and then we will go under the hood: the language pipeline (lexer, recursive-descent parser, AST, stack-based interpreter that yields one move per frame), the scoring that walks the AST, the text-based .lvl level format, and more. Also we will explain how the project was built using AI coding agents from specs and architecture I wrote, and what I would do differently.
Target audience:
developers and students interested in programming education, language design, software architecture and agentic development + geeks
RIVETS is a game that teaches programming to complete beginners: players write short programs in a tiny custom language to guide a robot through 15 maze levels. Reaching the goal is not enough: the game scores the player's code and rewards clean, abstracted solutions (loops and nesting) over brute-force sequences of moves, so players are nudged toward writing better code from level one. It was built with Java and Greenfoot as a Master's course assignment. The talk will include a live demo of the game and the level editor, and then we will go under the hood: the language pipeline (lexer, recursive-descent parser, AST, stack-based interpreter that yields one move per frame), the scoring that walks the AST, the text-based .lvl level format, and more. Also we will explain how the project was built using AI coding agents from specs and architecture I wrote, and what I would do differently.
Target audience:
developers and students interested in programming education, language design, software architecture and agentic development + geeks
Archontis E. Kostis
I’m Archontis Kostis, a software engineer and researcher from Thessaloniki, Greece. I wrote my first lines of code at 16, and what started as curiosity quickly turned into a passion. Today I’m pursuing my MSc in Software Engineering and Cloud Computing at the University of Macedonia. I also work as a researcher at the university’s SDE Lab, focusing on software quality and system design. There I build tools that measure code quality and show teams what to fix first.
I care about software that lasts, code that stays readable, maintainable and easy to change long after the first release.
I’m an active member of the local open-source community, and I believe that sharing knowledge openly makes all of us better engineers. When I’m not coding, I’m usually tinkering with hardware or reading.
I care about software that lasts, code that stays readable, maintainable and easy to change long after the first release.
I’m an active member of the local open-source community, and I believe that sharing knowledge openly makes all of us better engineers. When I’m not coding, I’m usually tinkering with hardware or reading.