fix(hidpp): keep events when a field carries an unknown enum value

[bugprone]

Summary

Event listener closures have no error channel, so a single unrecognised field byte made a whole event decode return None — a crown update with an unknown gesture, for example, lost its valid rotation deltas in the same payload. This sweeps the remaining event decoders onto the tolerant pattern #381 established.

The idiom is chosen per how each enum is used:

• Event-read-only enums (crown RotationState/ActivityState/CrownGesture/ButtonState, DpiDirection, BrightnessClampedSource, ActivityEventType, PowerModeTarget) gain a num_enum(catch_all) Other(u8) variant, so an unknown wire value carries its raw byte instead of dropping the event.
• Lod is also written back to the device via SetDpiParameters, so making it catch_all would let an invalid lift-off value be written. It stays a closed enum — keeping illegal values unrepresentable — and only its event field becomes Option<Lod>. The write and response paths stay strict: unknown values there still surface UnsupportedResponse.
• RawWheelResolution and IlluminationState are masked to their full value range at the call site and cannot fail, so they are left unchanged.

Unknown event sub-ids still return None: an unknown event kind has no payload shape to decode, unlike an unknown field within a known event.

Changes

• crates/openlogi-hidpp: tolerant decode for the crown, extended_dpi, illumination, and rgb_effects event decoders; Lod event field is now Option<Lod>.

Testing

• cargo test -p openlogi-hidpp — 161 passed. New totality tests sweep 0..=255 across each enum field position to prove a known sub-id never drops, plus the two prior "unknown → dropped" tests are flipped to assert the value is preserved. The strict response-path test (rejects_unknown_lod_value) stays green.
• cargo clippy -p openlogi-hidpp --all-targets and cargo fmt --check clean.
• Not runtime-tested on hardware — no physical device to exercise the unknown-value paths.

Fixes #382

[greptile-apps]

Greptile Summary

This PR applies the tolerant-decode pattern across the remaining HID++ event decoders so that a single unrecognised enum byte no longer silently drops a whole event. The strategy is consistently applied: enums used only in event reading gain a catch_all Other(u8) variant via FromPrimitive, while enums that are also written to the device (Lod, DpiDirection, PowerModeTarget) stay as closed TryFromPrimitive enums and their event fields become Option<_>.

• Crown (0x4600): RotationState, ActivityState, CrownGesture, and ButtonState all gain Other(u8); the decoder switches from try_from().ok()? to infallible from().
• ExtendedDpi (0x2202): DpiParametersChanged::lod and DpiCalibrationCompleted::direction become Option<Lod> / Option<DpiDirection>; write-path APIs remain strict.
• Illumination (0x1990) / RgbEffects (0x8071): BrightnessClampedSource and ActivityEventType get catch_all Other(u8); PowerModeTarget (packed into a 2-bit field) stays closed with the event field as Option<PowerModeTarget>. Each changed path is covered by a new 0–255 totality sweep test.

Confidence Score: 5/5

Safe to merge; the change tightens event-decode robustness without touching any request-encoding or write path.

All write-path APIs remain strict (Lod, DpiDirection, PowerModeTarget keep their closed TryFromPrimitive enums), the tolerant paths are gated behind event-only types, and every changed decode position is covered by a 0..=255 totality sweep. The only finding is a stale doc-comment in extended_dpi/event.rs — the code itself is correct.

The doc-comment on decode_event in crates/openlogi-hidpp/src/feature/extended_dpi/event.rs still describes the old drop-on-unknown-value behaviour and should be updated.

Important Files Changed

Filename	Overview
crates/openlogi-hidpp/src/feature/crown/event.rs	Converts four crown event enums from TryFromPrimitive to FromPrimitive with catch_all Other(u8); event decode path switches from ok()? to from(), eliminating all silent drops for sub-id 0.
crates/openlogi-hidpp/src/feature/crown/tests.rs	Renames the old dropped-unknown-enum test to confirm the event is preserved, and adds a full 0..=255 sweep across all five enum-field positions for sub-id 0.
crates/openlogi-hidpp/src/feature/extended_dpi/event.rs	Lod and DpiDirection event fields become Option<_> via try_from().ok(); the decode_event doc-comment still claims None is returned for unsupported enum values, which is now incorrect.
crates/openlogi-hidpp/src/feature/extended_dpi/tests.rs	Updates existing assertions to Some(Lod::Medium) / Some(DpiDirection::Y), flips the drop test, and adds a totality sweep over both enum-field positions for sub-ids 0 and 1.
crates/openlogi-hidpp/src/feature/illumination/event.rs	BrightnessClampedSource field now uses FromPrimitive::from() instead of try_from().ok()?, preserving the clamp event for any source byte.
crates/openlogi-hidpp/src/feature/illumination/tests.rs	Adds a full 0..=255 totality sweep for sub-id 4 (BrightnessClamped), asserting the brightness sibling field survives every source byte.
crates/openlogi-hidpp/src/feature/illumination/types.rs	BrightnessClampedSource gains FromPrimitive + catch_all Other(u8) variant; TryFromPrimitive retained for any write-path consumers.
crates/openlogi-hidpp/src/feature/rgb_effects/event.rs	UserActivity now uses ActivityEventType::from(); ClusterChanged power_mode becomes Option via try_from().ok(), addressing the previously missing ClusterChanged sweep.
crates/openlogi-hidpp/src/feature/rgb_effects/tests.rs	Adds two totality tests: a 0..=255 sweep for UserActivity (sub-id 1) and a full flags-byte sweep for ClusterChanged (sub-id 2), covering the gap noted in the prior review thread.
crates/openlogi-hidpp/src/feature/rgb_effects/types.rs	ActivityEventType converted to FromPrimitive with catch_all Other(u8); PowerModeTarget correctly left as a closed TryFromPrimitive enum since it lives in a 2-bit packed field.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Raw HID++ event payload] --> B{Known sub-id?}
    B -- No --> C[return None]
    B -- Yes --> D{Enum field type?}
    D -- catch_all FromPrimitive --> E[Known value gets concrete variant, unknown value gets Other byte]
    E --> F[Some Event always returned]
    D -- Closed TryFromPrimitive --> G{try_from succeeds?}
    G -- Yes --> H[field is Some variant]
    G -- No --> I[field is None]
    H --> J[Some Event with field Some]
    I --> K[Some Event with field None]
    F --> L[Caller receives event with all sibling fields intact]
    J --> L
    K --> L

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    A[Raw HID++ event payload] --> B{Known sub-id?}
    B -- No --> C[return None]
    B -- Yes --> D{Enum field type?}
    D -- catch_all FromPrimitive --> E[Known value gets concrete variant, unknown value gets Other byte]
    E --> F[Some Event always returned]
    D -- Closed TryFromPrimitive --> G{try_from succeeds?}
    G -- Yes --> H[field is Some variant]
    G -- No --> I[field is None]
    H --> J[Some Event with field Some]
    I --> K[Some Event with field None]
    F --> L[Caller receives event with all sibling fields intact]
    J --> L
    K --> L

Loading

_{Reviews (2): Last reviewed commit: "fix(hidpp): keep DpiDirection and PowerM..." | Re-trigger Greptile}