v0.5.6

Fixed

• LocalCommsStatus fields were mislabeled (previously guessed from limited
  captures). Corrected to the verified field semantics: field 2 health →
  status, field 3 link_state → visible_devices_count, field 4 version →
  expected_devices_count, field 5 health_changed_ts →
  last_session_change_ts, field 6 connection_state (int32) → reason
  (LocalCommsHealthReason enum). Decoding still worked by field number, but the
  model exposed wrong names/semantics and discarded the reason diagnostics.
  QuiltSmartModule/Controller now expose local_comms_visible_devices,
  local_comms_expected_devices, local_comms_reason, and
  local_comms_last_session_change (replacing local_comms_link_state,
  local_comms_connection_state, and local_comms_version).

Added

• IndoorUnit now exposes model_sku, serial_number, and firmware_version,
  resolved from HomeDatastoreSystem.indoor_unit_hardware (previously discarded).
  This mirrors OutdoorUnit and Controller and lets downstream consumers (e.g.
  the Home Assistant integration) populate IDU device serial/firmware.
• CLI set command gained a --fan option (AUTO, QUIET, LOW, MEDIUM,
  HIGH, BLAST) that applies the chosen fan speed to every indoor unit in the
  target space.
• SystemSnapshot.indoor_units_for_space() returns all indoor units linked to a
  space (accepts a Space or space-ID string).
• LocalCommsHealthReason enum (12 values). ZENOH_SESSION_DOWN
  identifies the local mesh transport as Eclipse Zenoh, not NATS (port 7447
  is Zenoh's default).
• IndoorUnitConditions.compressor_minimum_run_time (proto field 13).
• Marked SpaceSettings.safety_heating=9 confirmed (was "unconfirmed").

CI

• Docs deploy workflow upgraded to the current Node 24 GitHub Pages actions
  (configure-pages@v6, upload-pages-artifact@v4, deploy-pages@v5),
  removing the deprecated Node 20 runtime path.
• Docs deploy now retries deploy-pages once on transient GitHub Pages backend
  failures, failing the job only if both attempts fail.

v0.5.5

chore: bump version to 0.5.5

v0.5.4

Fixes all findings from a full architecture/code/performance/bug-hunt evaluation.

Critical

• Proto3 absence detection never fired on real protos. Truthiness and
  getattr-with-default checks always pass on generated messages (unset
  sub-messages return truthy default instances), so sparse stream diffs zeroed
  room state, IDU controls, sensor readings, and controller temperatures on
  merge. All model parsing now uses HasField-based helpers, and
  SystemSnapshot.apply_* preserves identity/relationship fields, controller
  temperatures, ODU telemetry, and QSM LED state. New test suite
  (tests/test_models_real_proto_merge.py) reproduces the notifier stream's
  serialize/parse path with real generated protos.
• Transport UNAUTHENTICATED retry was dead code. In grpc.aio, awaiting the
  interceptor continuation returns the call object; AioRpcError only surfaces
  when the call is awaited — so token refresh/retry never executed and clients
  broke permanently ~1h after token expiry. The interceptor now awaits the call
  and retries once. Tests updated to model real grpc.aio semantics.

High

• Stream reconnect backoff/budget reset after a healthy connection (routine LB
  stream recycling no longer escalates to permanent 60s delays or kills the
  stream after N lifetime disconnects); non-gRPC failures reconnect and surface
  via on_error instead of dying silently; malformed events are skipped;
  jitter added; prompt reconnect after successful token refresh.
• authenticate() no longer returns a locally-unexpired token the server just
  rejected; rotated Cognito refresh tokens are persisted; expiry measured from
  token receipt.
• grpc_call passes CancelledError through untranslated (asyncio.timeout/
  cancellation semantics restored).
• TUI: fatal stream death now shows a disconnect indicator and auto-recovers;
  first-time users get an OTP modal; boot failures show an error screen
  instead of a stuck spinner.

Medium / Low

• Unified error translation: all hds/user RPCs route through grpc_call;
  UNAVAILABLE/DEADLINE_EXCEEDED → QuiltConnectionError and NOT_FOUND →
  QuiltNotFoundError everywhere.
• Single-flight guards on token refresh and cold snapshot fetch;
  get_system_id(home=...) no longer poisons the default-system cache;
  close() stops tracked streams; duplicate authorization metadata
  eliminated.
• Stream API: on_connected callback, unsubscribe handles from all on_*
  registrations, apply_software_update_info, descriptor-derived wire-scan
  field numbers.
• CLI/TUI: app-owned snapshot (stacked screens no longer go stale), cursor
  preservation, setpoint bounds, energy period validation, app-level °C/°F,
  LED brightness restoration, 0.0 readings rendered correctly, token store
  corruption recovery + fsync + flock, config dir 0700, dead code removal.
• Docs: README no longer demonstrates raw-stream usage; stream/token types
  re-exported at package top level; stale OutdoorUnit/Controller reference
  listings corrected; contradictory proto comments fixed.

Intentionally unchanged: Python >= 3.14 floor and boto3 dependency (HA 2026.3+
runs Python 3.14).

v0.5.3

chore: bump version to 0.5.3

v0.5.1

Fixed

• SpaceControls.display_setpoint_str() — removed dead unreachable code in the
  fallback branch; temperature_setpoint_c is typed float so the None-guard
  lines were never executed (confirmed by coverage)
• SystemSnapshot.apply_outdoor_unit() — refactored to collect field patches into
  an updates dict and call dataclasses.replace() once, consistent with all other
  apply_* methods; previously called replace() twice in sequence, creating an
  unnecessary intermediate object
• QuiltClient.close() now sets self._token = None; previously left a stale token
  accessible via get_current_token() after the channel was closed

Changed

• invoke_refresh_callback (formerly _invoke_refresh_callback) extracted from
  transport.py and services/streaming.py into a single shared implementation in
  tokens.py; the streaming copy lacked the WeakKeyDictionary signature cache,
  causing inspect.signature() to be called on every token-refresh event
• FanSpeed.to_wire() and LouverAngle.to_wire() now reference module-level
  constant dicts (_FAN_SPEED_WIRE_MAP, _LOUVER_ANGLE_WIRE_MAP) instead of
  re-allocating the mapping on every call
• _id_variants() moved from models/system.py into models/_helpers.py and
  reused by lookup_hardware(); eliminates duplicated ID-normalisation logic
• QuiltClient.invalidate_snapshot() log level changed from WARNING to DEBUG

v0.5.0

Added protocol support

• HVACMode.DRY = 8 — dehumidification mode; gate: mobile_dry_mode_selection_enabled. Fan non-interactive (QSM forces ~600 RPM). No user-configurable temperature setpoint; built-in temperature floor is server-side.
• HVACState.DRY = 11, DRY_DEFERRED = 12, DRY_PREPARING = 13
• LocalCommsHealthStatus enum (UNSPECIFIED=0, HEALTHY=1, DEGRADED=2, OFFLINE=3, STARTING_UP=4) — gate: mobile_local_control_health_enabled
• QuiltSmartModule.local_comms_health — extracted from new LocalCommsStatus nested message (proto field 8, subfield 2)
• Controller.local_comms_health — extracted from new LocalCommsStatus nested message (proto field 9, subfield 2)
• LocalCommsStatus proto message with updated_ts, health, link_state, version, health_changed_ts, connection_state subfields (fields 1–6; wire-confirmed)

Changed

• APP_VERSION bumped to 1.0.26
• LocalCommsStatus is a nested message (not a simple enum) on both QSM and Controller — the earlier inferred field type was incorrect; wire-confirmed via 2026-06-04 mitmproxy capture

v0.4.0

chore: bump version to 0.4.0

v0.3.2

Added

• LouverAngle.label property and __str__ with human-readable position names:
  ANGLE1 → "Horizontal", ANGLE2 → "Slightly Down", ANGLE3 → "Down",
  ANGLE4 → "Mostly Down", ANGLE5 → "Straight Down"

v0.3.1

Fixed

• RST_STREAM with error code 0 (HTTP/2 NO_ERROR, a normal server-side graceful reset) is now logged at DEBUG instead of WARNING to reduce log noise
• CANCELLED (server closed the stream normally, e.g. keepalive timeout or server rotation) is now logged at INFO instead of WARNING
• UNAUTHENTICATED reconnects handled by the automatic token refresh are now logged at INFO instead of WARNING

v0.3.0

Added

• NotifierStream health properties: is_connected, last_event_at, stream_state
• NotifierStream debounce_s parameter to coalesce rapid update bursts
• MetricBucketStatus enum exposed on EnergyBucket.status (was untyped int)
• grpc_call() context manager in services for consistent gRPC error translation and optional retry
• Structured logging.getLogger(__name__) across all modules (auth, client, transport, services, CLI)
• QuiltStreamError re-exported from the top-level quilt_hp package
• Shared model helpers (_helpers.py) for WiFi signal parsing and hardware lookup

Changed

• ScheduleEvent.hvac_mode is now typed as HVACMode (was int)
• Token temp file created with os.open(..., 0o600) so permissions are secure from creation (no transient world-readable window)
• Signature cache in transport layer uses weakref.WeakKeyDictionary instead of dict[int, bool] — prevents unbounded growth and id-reuse bugs after GC
• login() clears the token cache so a re-login always fetches fresh credentials
• _GrpcCallContext avoids self-chaining when re-raising a QuiltError unchanged

Fixed

• EnergyBucket.has_missing_energy_value now treats None (absent proto field) as missing, not just NaN; prevents TypeError in SpaceEnergyMetrics.total_kwh
• MetricBucketStatus() conversion in get_energy_metrics catches ValueError for unknown server values and falls back to UNSPECIFIED instead of raising
• WeakKeyDictionary.get() for the refresh-callback signature cache is now guarded against TypeError for non-weakrefable callables
• AUTO mode setpoint deadband clamp now runs before setpoint selection, ensuring the correct (clamped) value is sent to the device
• CLI settings bool coercion uses isinstance(v, bool) to avoid bool("false") == True
• Zero-value proto3 fields (0 °C temperature, 0 dBm WiFi signal, 0% humidity) are now preserved instead of being dropped as falsy
• NotifierStream reconnect subscription is now protected by an asyncio.Lock to prevent concurrent subscribe/reconnect races
• CLI enum lookups raise a clear error showing valid options on invalid input
• auth.py narrows broad except Exception to except (QuiltAuthError, ClientError) to avoid swallowing unexpected errors