OpenRecomp MIPS32 — CANDIDATE to PROVEN

Aim

Help harden OpenRecomp's existing MIPS32 static recompilation path from CANDIDATE toward PROVEN through reproducible, public evidence.

OpenRecomp is an open-source project exploring deterministic static recompilation as a tool for software preservation.

Static recompilation translates software built for one computer architecture into native code that can run on modern systems. OpenRecomp is being developed around a simple principle:

Preserve. Understand. Recompile.

What already works

OpenRecomp already has a PROVEN RV32I synthetic baseline, demonstrating a bounded and reproducible static-recompilation pipeline.

MIPS32 is the project's second architecture. It is not starting from zero. Bounded MIPS32 fixtures and validation paths already exist and pass their documented tests.

Those results do not establish compatibility with arbitrary MIPS32 software, so wider MIPS32 support remains:

CANDIDATE

This campaign will fund the engineering and validation needed to harden that existing work and attempt to move the documented MIPS32 scope toward PROVEN.

What will the funding pay for?

The £10,000 net engineering programme is divided into six milestones:

£2,000 — Ingestion and extraction hardening
Strengthen the path from input binaries to deterministic program representation.

£3,000 — Semantics and translation
Expand and validate the bounded MIPS32 instruction and translation corpus, including little- and big-endian cases.

£1,500 — Indirect control flow
Improve bounded handling and evidence for jump tables and indirect control-flow cases.

£1,500 — Native and WebAssembly validation
Strengthen parity, repeatability and deterministic-output testing across supported host paths.

£1,000 — Rights-safe test corpus
Expand synthetic, open or otherwise authorised fixtures, including negative tests and reproducibility evidence.

£1,000 — Documentation and public evidence
Publish results, limitations, reproducibility material and the final evidence classification.

The Crowdfunder target also accounts for the transaction fees charged by the platform.

Funding does not guarantee PROVEN

An important principle of OpenRecomp is that funding does not determine the result — evidence does.

This campaign funds the work required to test and strengthen the MIPS32 implementation.

If the required validation succeeds, the documented scope can move toward PROVEN. If it does not, the failures and remaining limitations will be documented publicly instead.

Existing work will not be rebilled as new campaign work.

Open, reproducible and rights-safe

The work will be developed openly, with reproducible tests and public evidence wherever practical.

The campaign does not fund distribution of copyrighted games, proprietary firmware, console keys or proprietary SDK material.

Validation will use synthetic, open, homebrew, public-domain or otherwise authorised material as appropriate.

OpenRecomp is currently maintained primarily by one developer. Public source code, reproducible evidence, documentation and milestone artefacts are intended to reduce that dependency and make the work independently inspectable and continuable.

Why support OpenRecomp?

Software preservation increasingly depends on understanding programs whose original hardware and operating environments are disappearing.

OpenRecomp is exploring whether deterministic static recompilation can become another practical, inspectable tool for that preservation work.

This campaign is a deliberately bounded next step: take an existing MIPS32 CANDIDATE, harden it, test it, document the results and let the evidence determine what can legitimately be called PROVEN.

Preserve. Understand. Recompile. Keep software running.

Back OpenRecomp MIPS32 — help harden CANDIDATE toward PROVEN.

Funding method

All or nothing – this project will only receive pledges if it reaches its target of £10,500 by 16th October 2026 at 2:49pm