Skip to content

Unmapped clusters on an endpoint with a recognised device are dropped silently (no log, no placeholder) #81

Description

@simons-plugins

Problem

_unknown_spec() (and its INFO log + matterUnknown placeholder) only fires when an endpoint produces no handler specs at all. If any handler claims the endpoint (e.g. OnOff → matterRelay), every other unmapped cluster on that same endpoint is dropped with zero diagnostic — worse than the #79 symptom, which at least produced a visible placeholder.

Concrete example (already in the tracker)

#70's PumpConfigurationAndControl (0x0200) is co-located with OnOff on pump endpoints: a Matter pool pump today creates a working relay and silently loses all pump-speed capability, with nothing in the log inviting a device-support report.

Also affected when they ship: Water Heater Management (0x0094) and Device Energy Management (0x0098/0x009F), which typically co-locate with recognised clusters.

Suggested behaviour

In create_devices(), after specs are planned for an endpoint, compute the endpoint's leftover clusters (not claimed by any produced spec's handler, not in _NON_DEVICE_CLUSTERS) and emit the same style of INFO log as the placeholder path ("endpoint N also exposes clusters this plugin does not support yet (…) — report at …"), without creating a second device. Zoo fixture: mixed_unmapped (OnOff + 0x0200 on one endpoint) pinning that the log fires.

Found by the 2026-07-05 cluster-coverage audit (bug class of #79 / #58). Being fixed in PR #80.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Arj5R66d73tc3PMPvHS65Z

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions