July 10, 2026 · 6 min read
Modernizing Cyberline Racing: five Unity versions and a new backend
How we took a vehicular-combat racing game from a decade-old Unity build with a dead backend to a modern, maintainable codebase.
Cyberline Racing is a vehicular-combat racing game for Android, originally developed by Magicindie Softworks. aIPc Soft now maintains and publishes it, and getting it back into shape was a serious modernization project. Here is what that work actually involves, because it is a common situation: a good product built on a technology stack that time has left behind.
The starting point
The game arrived as a large Unity project: hundreds of gameplay scripts, nine racing tracks, a garage-and-upgrades economy, in-app purchases, rewarded ads, and cloud save. Mechanically it was solid and fun. Technically, it was stuck: it was built on a very old version of Unity, and its online backend ran on a service (App42) that had effectively gone dark. Cloud saves and multiplayer-adjacent features had nothing to talk to.
You cannot ship that to a modern app store, and you cannot patch around it. It needed real modernization work.
Climbing the Unity ladder
You generally cannot jump a Unity project across many major versions in one leap: APIs are removed, rendering changes, plugins break. So we moved up the ladder one supported version at a time: Unity 5 → 2017 → 2018 → 2019 → 2020.3, with the goal of reaching a current LTS release before store publication.
At each step we fixed what that version broke, verified the game still ran, and only then moved on. This is slower than it sounds like it should be, but it is the only reliable way to keep a large project working across a decade of engine changes. Throughout, we validated on a real device rather than trusting the editor.
Replacing a dead backend
The old App42 backend was not coming back, so we built a new one from scratch on a modern, boring, dependable stack: Java 21, Spring Boot 3.3, and MongoDB. This handles the online-facing pieces the game needs, on infrastructure we control and can maintain. The new backend went live in July 2026.
The lesson
Legacy does not mean worthless. A lot of valuable software is trapped on old runtimes: games, internal tools, apps that customers still use. The work is unglamorous: migrate carefully, verify constantly, and replace the parts that have rotted with something you can actually maintain.
If you have a product in this situation, one that works but is stuck on old technology, that is exactly the kind of problem we take on.