safety-critical-patterns

Safety-Critical Coding Patterns

Guidelines adapted from NASA's Power of 10 rules for safety-critical software.

When to Apply

Full rigor: Safety-critical systems, financial transactions, data integrity code Selective application: Business logic, API handlers, core algorithms Light touch: Scripts, prototypes, non-critical utilities

"Match rigor to consequence" - The real engineering principle

When NOT To Use

• Ordinary application code, where these defensive checks become the bloat that prefer-invariants-over-fallbacks targets (use conserve:code-quality-principles)

The 10 Rules (Adapted)

1. Restrict Control Flow

Avoid goto, setjmp/longjmp, and limit recursion.

Why: Ensures acyclic call graphs that tools can verify. Adaptation: Recursion acceptable with provable termination (tail recursion, bounded depth).

2. Fixed Loop Bounds

All loops should have verifiable upper bounds.

# Good - bound is clear
for i in range(min(len(items), MAX_ITEMS)):
    process(item)

# Risky - unbounded
while not_done:  # When does this end?
    process_next()

Adaptation: Document expected bounds; add safety limits on potentially unbounded loops.

3. No Dynamic Memory After Initialization

Avoid heap allocation in critical paths after startup.

Why: Prevents allocation failures at runtime. Adaptation: Pre-allocate pools; use object reuse patterns in hot paths.

4. Function Length ~60 Lines

Functions should fit on one screen/page.

Why: Cognitive limits on comprehension remain valid. Adaptation: Flexible for declarative code; strict for complex logic.

5. Assertion Density

Include defensive assertions documenting expectations.

def transfer_funds(from_acct, to_acct, amount):
    assert from_acct != to_acct, "Cannot transfer to same account"
    assert amount > 0, "Transfer amount must be positive"
    assert from_acct.balance >= amount, "Insufficient funds"
    # ... implementation

Adaptation: Focus on boundary conditions and invariants, not arbitrary quotas.

6. Minimal Variable Scope

Declare variables at narrowest possible scope.

# Good - scoped tightly
for item in items:
    total = calculate(item)  # Only exists in loop
    results.append(total)

# Avoid - unnecessarily broad
total = 0  # Why is this outside?
for item in items:
    total = calculate(item)
    results.append(total)