MM Compiler

A compiler that adds zero bytes. Everything you see is earned through barcode scans.

The Gate

The only hardcoded logic in the entire application:

function canAccept(prime, generated) {
  // Direct: this number exists in the generated set
  if (generated.has(prime)) return true;

  // GCD: shares a factor with anything generated
  for (const existing of generated) {
    if (gcd(prime, existing) > 1) return true;
  }

  return false;
}

function gcd(a, b) {
  a = Math.abs(a); b = Math.abs(b);
  while (b) { [a, b] = [b, a % b]; }
  return a;
}

Everything else — CSS, content, fonts, database tables, the lexicon — arrives through barcodes and must pass this gate.

The Ledger (mm_loop)

The heart of the system. A single table in localStorage:

When you scan a raw prime:

1. The prime is added with cost = φ²
2. Every product (prime × existing values) is added as a composite with cost = 0
3. The pending queue is swept — any barcode whose prime now passes the gate gets replayed

The potential is: SELECT * FROM mm_loop WHERE consumed = false

Barcode Protocol

Format: type:prime:content

What the IDE Builds

1. The gate function (canAccept + gcd)
2. The ledger table (mm_loop)
3. The scan input (camera + manual + calculator)
4. The DB visualizer (Studio tab)

That's it. Everything else arrives through barcodes.

What the IDE Does NOT Build

• ❌ No CSS (earned from css: barcodes)
• ❌ No fonts (earned)
• ❌ No content (earned from h1:, p:, etc.)
• ❌ No database engine (earned from js: barcodes)
• ❌ No lexicon (earned from lex: barcodes)

Boot Sequence

1. Phase 1: Scan raw primes (2, 3, 5, 7, 11, 13, 17, 19) — build the ledger
2. Phase 2: Scan CSS tensors — the app becomes beautiful
3. Phase 3: Scan content — the paper materializes
4. Phase 4: Scan JS — database engine arrives
5. Phase 5: Scan lexicon — the app speaks English

The compiler adds zero bytes of content. Zero bytes of styling. Zero lines of business logic. It only knows how to accept or reject. Everything else is earned.