Arbitrary Models vs Raw Bytes

This post is part of the Input Coverage > Code Coverage series.

Two strategies:

• Share Arbitrary models. Great for domain logic. Faster feedback.
• Use raw &[u8]. Great for broad input coverage. Fewer blinders.

Heuristics:

• If you already validated structure elsewhere, fuzz raw bytes.
• If structure is complex and slow to derive, use an Arbitrary model to prune junk, but keep a raw-bytes harness too.

Watch discard rates. If most generated values are rejected, simplify.

---

Next: Mocking I/O To Go Faster