# Changelog

## Score

- CAI 73 → 74 (+1.7)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

## Lenses

- Code Health 97 → 97 (+0.0)
- Architecture 88 → 88 (+0.0)
- Maturity 57 → 61 (+3.7)
- Readiness 80 → 80 (-0.1)
- Security 100 → 100 (+0.5)
- Event Sourcing 100 → 100 (+0.0)
- Performance 90 → 90 (+0.0)

## Resolved (4)

- Confusing naming convention. `process_poll_event` (singular) vs `process_poll_events_until_timeout` (plural + duration) is inconsistent. It is unclear if the singular version processes one event and blocks, or processes all available events. The naming does not clearly convey the iteration behavior.
- Inconsistent async handling. `for_each` implies a synchronous iteration, but `for_each_async` returns a `NodeTask`. This suggests `for_each` might be blocking or synchronous, while the async version is non-blocking. However, in an async runtime, having a synchronous `for_each` that blocks the thread is often an anti-pattern or requires careful documentation. The naming suggests they are variants of the same operation, but the signatures are fundamentally different in how they handle execution.
- Redundant method naming and parameter duplication. The distinction between `connect` and `connect_with` is unclear without documentation, as is the distinction between the base `Transport` and `TransportConnect`. Similarly, `connect` vs `connect_sync` implies a synchronous/asynchronous difference, but the API surface suggests `connect` might be async and `connect_sync` sync, yet both return `Result`. This creates a confusing combinatorial explosion of methods for a single operation.
- Redundant method naming. Similar to connect, the distinction between `listen` and `listen_with` is ambiguous. Why does one take `Transport` and the other `TransportListen`? If `TransportListen` is a specific config, the naming should reflect that (e.g., `listen_with_config`), or the base method should accept the config.

## New (1)

- Outdated: log
