For each public method, identify the hardest-to-test edge cases:
- Null values (where permitted)
- Maximum/minimum weight
- Zero-duration expiration
- Mapping functions that return the same instance
- Mapping functions that throw
- Weighers that throw
- Expiry callbacks that throw
- CompletableFuture values (async cache edge cases)
- Keys/values with adversarial hashCode/equals
For each edge case, trace the code path. Does the code handle it correctly?
Identify combinatorially hard behavioral dimensions:
- Operation A on expired entry during concurrent operation B
- Exception in user callback X while holding lock Y
- GC collecting reference R between code points P and Q
- Fast path falls through to slow path under contention (e.g., entry appears expired on fast path, recovers under lock on slow path)
- Async cache: future completes between check and action (e.g., isComputingAsync returns true, but future completes before the code that depends on that check executes)
For each candidate gap, provide a minimal test case with the specific cache configuration and thread interleaving needed to reach the code path.

Priority ordering:

Paths involving the catch-commit-rethrow pattern (doComputeIfAbsent, remap)
Slow paths reachable only via contention (synchronized blocks after optimistic checks)
Interactions between expiration and the async value lifecycle (ASYNC_EXPIRY, isComputingAsync)

Focus only on behavioral coverage gaps that could hide correctness bugs.

المزيد من هذا المستودع

نفس المستودع

audit-adaptivity

ben-manes/caffeine

Audit the adaptive window hill-climber and region-resize logic for implementation defects (not algorithm quality)

2026-06-1717.7k

audit-jcache-conformance

ben-manes/caffeine

JSR-107 (JCache) spec-conformance audit

2026-06-1717.7k

audit-state-machine

ben-manes/caffeine

Audit explicit state machines (drain status, node lifecycle, async-value lifecycle) for illegal or missed transitions

2026-06-1717.7k

audit-temporal-walk

ben-manes/caffeine

Heavyweight history-mining bug audit. Walks the caffeine module's git history chronologically (oldest to HEAD), maintains a forward-tracked issue database, and surfaces concerns introduced by past commits that were never resolved. Catches bugs that snapshot mining cannot — half-fixes invisible from current state, latent+trigger pairs across multi-commit interactions, and partial refactors. Slow (model/effort-dependent; ~24h on Opus + max effort) and rare-run (every several months or before a major release).

2026-06-1717.7k

audit-sibling-divergence

ben-manes/caffeine

Differential audit comparing matched code paths that should behave identically. Spawns one auditor per sibling pair (sync/async, bounded/unbounded, view consistency, bulk vs single, generated node variants, read fast vs slow, adapter conformance) and requires a concrete witness scenario where the two paths diverge observably.

2026-06-0217.7k

audit-contract-drift

ben-manes/caffeine

Find places where documented API contracts and the implementation diverge

2026-04-2717.7k

name	audit-coverage-gaps
description	Discover test coverage gaps that could hide correctness defects
context	fork
agent	auditor
disable-model-invocation	true

Assume the existing tests miss at least one real defect.

For each public method, identify the hardest-to-test edge cases:
- Null values (where permitted)
- Maximum/minimum weight
- Zero-duration expiration
- Mapping functions that return the same instance
- Mapping functions that throw
- Weighers that throw
- Expiry callbacks that throw
- CompletableFuture values (async cache edge cases)
- Keys/values with adversarial hashCode/equals
For each edge case, trace the code path. Does the code handle it correctly?
Identify combinatorially hard behavioral dimensions:
- Operation A on expired entry during concurrent operation B
- Exception in user callback X while holding lock Y
- GC collecting reference R between code points P and Q
- Fast path falls through to slow path under contention (e.g., entry appears expired on fast path, recovers under lock on slow path)
- Async cache: future completes between check and action (e.g., isComputingAsync returns true, but future completes before the code that depends on that check executes)
For each candidate gap, provide a minimal test case with the specific cache configuration and thread interleaving needed to reach the code path.

Priority ordering:

Paths involving the catch-commit-rethrow pattern (doComputeIfAbsent, remap)
Slow paths reachable only via contention (synchronized blocks after optimistic checks)
Interactions between expiration and the async value lifecycle (ASYNC_EXPIRY, isComputingAsync)

Focus only on behavioral coverage gaps that could hide correctness bugs.