NanyangOJ: design and implementation of a participant-facing online judge platform for competitive programming

Summarise with VIRAT
Authors: Yutong Tang
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Competitive programming contests at universities rely on online judge systems to accept, compile, and evaluate participant code under strict resource limits. Existing platforms present a trade-off that none resolves cleanly: DOMjudge offers solid open-source judging infrastructure but ships a participant interface largely unchanged since the mid-2000s; Kattis provides a polished web experience but cannot be self-hosted without a commercial licence; Codeforces supports millions of users but offers no mechanism for private, institutionally controlled contests. NTU’s College of Computing and Data Science needed a system that combined self-hosting, modern browser-based code editing, and real-time contest communication — and none of these three options delivered all three. This report describes the design, implementation, and evaluation of the participant-facing frontend for NanyangOJ, a full-stack competitive programming platform built internally at NTU CCDS. We built the frontend as a React 19.1.0 single-page application in TypeScript 5.8.3, styled with Material-UI 7.2.0 and Tailwind CSS 4.1.10, and bundled with Vite 8.0.0. The interface embeds Microsoft’s Monaco Editor via @monaco-editor/react 4.7.0, giving participants a VS Code-grade coding environment directly in the browser with support for C, C++, Java, Python, and Go. Jury clarifications reach participants in real time through a Server-Sent Events stream, bypassing the polling-based delivery that older systems rely on. Every API response passes through Zod 4.0.5 runtime schema validation before reaching the UI, catching backend contract changes at the integration boundary rather than letting them propagate as silent rendering failures. We evaluated the system across eight functional test scenarios covering the full contest lifecycle. All scenarios passed. The SSE stream delivered clarifications within approximately one second in local testing, and the Axios-based rate-limit auto-retry mechanism absorbed HTTP 429 responses transparently. The main limitation — no automatic SSE reconnection after a network interruption — is documented and a fix is outlined for future work.

Fields of Study & Subjects

Mobile and Web Applications Teaching and Learning Programming Engineering and Information Technology

Publication details

Journal / source
DR-NTU (Nanyang Technological University)
Publication date
1 January 2026
Language
EN