Evidence at a glance #
Based on the linked sources. The author's reported result and independent confirmation are separate checks.
- AI involvement
- Reported models: Claude. Tools: PyBoy, Mesen 2.1.1, MAME 0.288, Python, Node.js. The writeup below describes the work attributed to AI.
- Reported result
- Research. This is the directory's source-based outcome classification.
- Independent confirmation
- Not reported.
- Human work and limitations
- The project uses a rigorous oracle-based testing framework to ensure the JavaScript ports match the original game logic. It explicitly distinguishes between 'emulation' and 'translation'.
Original sources
What was done #
Mixup is a project that translates console game machine code into JavaScript. It uses an oracle-based validation method to ensure fidelity against original hardware, supporting Batman: Return of the Joker, Gradius, and DoDonPachi DaiOuJou.
Mixup is a research project focused on hand-translating console game disassembly into readable, modifiable JavaScript, explicitly avoiding emulation. The project maintains fidelity by using an oracle method: an emulator runs the original ROM to generate a per-frame state vector, which is then compared against the JavaScript port. The author reports that this method identifies bugs that manual disassembly reading misses, noting that downstream symptoms are not the bug. The project includes a strict asset-integrity gate to ensure no ROM-derived material is committed to the repository. Recent development, as documented in the commit history, involves porting DoDonPachi DaiOuJou, with Claude Code listed as a co-author on multiple commits. The project acknowledges that tests can remain green despite bugs, citing four specific failure modes in check design. Limits include the requirement for users to provide their own legally owned ROMs for asset extraction, as the repository contains no ROMs or extracted graphics. The project is currently in a research state with varying levels of completion across the three supported games.
Reproduce it #
Export assets from legally owned cartridges using the provided Python scripts, then run 'python -m http.server 8000' to serve the project locally.