Refs #40
File: crates/charon-core/src/profit.rs (tests module)
PR branch: feat/15-profit-calc-and-queue
Problem:
The workspace deny(arithmetic_side_effects, cast_possible_truncation) lints require all arithmetic to be explicitly checked. The PR claims checked_mul is used, satisfying the lint. However the 5 unit tests use small, well-behaved values and do not exercise:
- repay_amount_cents near u64::MAX (checked_mul returns None — verify error path)
- liquidation_bonus_bps == 10_000 (max valid; gross == 2x repay — borderline overflow check)
- liquidation_bonus_bps == 10_001 (invalid — verify rejection fires correctly)
- flash_fee + gas + slippage > gross (net goes negative in signed arithmetic or wraps in unsigned — verify the error variant is Overflow not BelowMinThreshold)
- All costs == gross exactly (net == 0 — boundary between rejection and acceptance)
- min_profit_usd_cents == 0 (should every positive net profit pass?)
These cases are especially important because calculate_profit is a safety gate. A test suite that passes on happy-path values gives false confidence.
Fix: Add proptest or table-driven tests covering the above boundary values. For the negative-net case verify the exact error variant returned. For u64::MAX repay verify checked_mul returns Overflow not a panic.
Refs #40
File: crates/charon-core/src/profit.rs (tests module)
PR branch: feat/15-profit-calc-and-queue
Problem:
The workspace deny(arithmetic_side_effects, cast_possible_truncation) lints require all arithmetic to be explicitly checked. The PR claims checked_mul is used, satisfying the lint. However the 5 unit tests use small, well-behaved values and do not exercise:
These cases are especially important because calculate_profit is a safety gate. A test suite that passes on happy-path values gives false confidence.
Fix: Add proptest or table-driven tests covering the above boundary values. For the negative-net case verify the exact error variant returned. For u64::MAX repay verify checked_mul returns Overflow not a panic.