Verify in-game: DualSense BT fused orientation no longer swings side-to-side on macOS (#83 / PR #84 sign-off)

[petegordon]

Type: Verification / acceptance sign-off · Parent: #83 · Verifies PR: #84 · Playtest target: petegordon/Tandemonium

Why

#83 was filed from an in-game observation: DualSense over Bluetooth on macOS → fused orientation swings/jerks hard side to side during petegordon/Tandemonium play. PR #84 hardens the IMU offset selection (accel-as-gyro guard). But hardware testing on branch claude/issue-83-macos-bt-imu-offset shows this machine does not reproduce the offset misread — the probe cleanly picks the correct offset every time:

IMU offset=16 (default=16, probe-best=16, accelMag≈8089, gyroAbs≈14) → family='sony-ds5'
Gyro calibrated, bias: … stddev: 1.1

So the lab/console evidence says parsing is correct (gyro ≈ 0, accel ≈ 1g). That means #84 may be confirming a no-regression rather than repairing the live symptom — and the only decisive test left is playing the game and watching the model.

What to do

1. Run petegordon/Tandemonium with the PlayStation BT IMU: guard against accel-as-gyro offset shift (macOS side-to-side) (#83) #84 build, DualSense connected over Bluetooth on macOS.
2. Steer / move the controller and watch fused orientation. Compare against the same controller over USB (the known-good baseline from macOS WebHID: DualSense Bluetooth gyro fusion produces wrong orientation (USB-vs-BT report layout) #83).
3. Try to provoke the original failure mode specifically: move/handle the pad during the ~600ms connect-time probe, hot-plug/re-pair a few times. The original bug required the probe tie-breaker to flip to a wrong neighbour (15/17); PlayStation BT IMU: guard against accel-as-gyro offset shift (macOS side-to-side) (#83) #84's gyro gate should now prevent that. Confirm offset stays 16 across repeated reconnects (watch the PlayStation IMU offset=… log line).

Two possible outcomes (record which)

• Orientation tracks cleanly, no side-to-side, across reconnects → macOS WebHID: DualSense Bluetooth gyro fusion produces wrong orientation (USB-vs-BT report layout) #83 is resolved by PlayStation BT IMU: guard against accel-as-gyro offset shift (macOS side-to-side) (#83) #84 (the original cause was the intermittent probe-flip, now guarded). Close macOS WebHID: DualSense Bluetooth gyro fusion produces wrong orientation (USB-vs-BT report layout) #83, merge PlayStation BT IMU: guard against accel-as-gyro offset shift (macOS side-to-side) (#83) #84.
• Side-to-side persists even with offset=16 logged → the residual cause is downstream in fusion (BT axis mapping / sign / scale), not parsing. Hand off to the sibling fusion-axis-mapping issue and keep macOS WebHID: DualSense Bluetooth gyro fusion produces wrong orientation (USB-vs-BT report layout) #83 open scoped to that.

Acceptance

• Documented playtest result (BT vs USB) with the PlayStation IMU offset=… log captured across ≥3 reconnects, at least one with the pad in motion during probe.
• Clear verdict: resolved by PlayStation BT IMU: guard against accel-as-gyro offset shift (macOS side-to-side) (#83) #84, or escalated to the fusion-axis issue.