Python testing: inject a clock to test expiry at the exact boundary

Operation contract

The lease is valid strictly before its expiry. The fixture passes the current instant as an integer rather than patching a global clock. This makes the equal-to-expiry case explicit, repeatable and independent of machine timezone or test execution speed. The same function rejects Boolean times and malformed lease fields before checking expiry. A test need not sleep to reach a boundary it can supply as data.

Failure and ownership boundary

An injected clock does not solve clock trust across hosts. Stored leases still need a deployed authority for timestamps, monotonic versus wall-clock policy and transaction ordering. The fixture checks a pure predicate only. Python SQLite lease renewal: reject stale tokens and exact-expiry ownership and Python expiry exercise: separate pending, active and expired records at exact boundaries connect the predicate to state transitions.

Working program

def active_lease(now_ms, expiry_ms):
    if type(now_ms) is not int or type(expiry_ms) is not int or not 0 <= now_ms <= 100000 or not 0 <= expiry_ms <= 100000:
        raise ValueError("bounded millisecond times")
    return now_ms < expiry_ms

for controlled_now in (499, 500, 501):
    print(controlled_now, active_lease(controlled_now, 500))
try:
    active_lease(True, 500)
except ValueError:
    print("boolean clock rejected")

Output

499 True
500 False
501 False
boolean clock rejected

Costs and limits

The predicate uses bounded integer comparisons and no sleeping. A production storage check has database and contention costs that are absent here. Time units and clock authority must remain part of the caller contract.

Common Mistakes

• A sleeping test can miss or delay the exact equality boundary.
• A local injected clock does not establish distributed clock agreement.

Connected lessons

Python expiry exercise: separate pending, active and expired records at exact boundaries, Python SQLite lease renewal: reject stale tokens and exact-expiry ownership, Pytest fixtures: isolated files and parameterized rejection tests.