Repository navigation
Validate final-image buttons, menus, reset stability, and audio cues on hardware #69
Description
Original request
Relevant excerpt, lightly grammar-corrected for publication. Credentials, personal data, private paths, and unrelated transcript content are omitted.
Physical front-panel keys, PC-sent virtual buttons, and RF-received keys must always respond immediately; fix the root cause and protect against regression.
Source: session edge-current, authored turn 204.
Parent epic: #13
Requirement
Validate final-image buttons, menus, reset stability, and audio cues on hardware
Acceptance criteria
- Test Buttons 1-4 Down/Up/Click, double-click, hold acceleration, and editors while monitoring reset count.
- Validate the nested settings fields, save/discard, default-page behavior, and no menu-navigation reset.
- Listen for boot melody, one clean beep per key, and save/discard/door/relay cues with Silent off.
- Confirm the first-run TUI remains synchronized through board ready and welcome-melody completion.
- Request this physical pass only after safe setup by showing WAIT and playing the unique repeating ringtone; stop the cue immediately after acknowledgement.
Current evidence and gaps
Earlier melody and buttons 1/2 evidence exists, but the final-image full key/menu/audio and synchronized first-run pass is not complete.
Traceability
- Canonical checklist section: Final hardware validation and handoff
- Includes only curated, issue-relevant prompt excerpts; complete transcripts, credentials, secrets, and private paths remain unpublished.
Activity
[-]Validate final-image buttons, menus, reset stability, and audio cues on hardware[/-][+]Validate final-image buttons, menus, reset stability, and buzzer cues on hardware[/+]atomicdeploy commentedon Aug 6, 2026
Production-board hardware run (third/live board):
atomicdeploy commentedon Aug 6, 2026
Fixed the repeated startup-melody / missing
Proglifecycle regression in df5c6e9.Root cause:
--reinitialize-eepromintentionally preserved the original raw EEPROM but never armed the durable programming latch afterward, then programmed a normal factory image with Silent/Programming clear. Urclock verification resets therefore looked like ordinary boots.The guarded lifecycle now:
Progbrightness;Live acceptance on COM5: fresh full backup
pccontroller-20260806-143218; latch changed false -> true; program mode became 0; one deliberate DTR reset advanced reset count exactly 10 -> 11; latch remained true after reboot; original settings were restored and normal sensing resumed. Current uptime evidence also disproved an ongoing reset loop.Validation:
go test ./...,go vet ./..., and 76 build/firmware contract tests pass.atomicdeploy commentedon Aug 6, 2026
Final merge/install evidence (2026-08-06): PR #113 is merged to
mainas87537eb897269fad717ae8a6d836b135910d0f8a. The merged-main Windows host was rebuilt and installed at the canonical AppData path. A fresh COM5 query through that installed binary reported firmware45E57A94, uptime1h45m54.317s, reset count11, settingsflags=0x01,programming_latch=false, and all relay/PWM outputs off. The reset count remains exactly the value reached by the one deliberate latch-persistence reset, so the repeated-reset/startup-melody regression is no longer present. The programming transaction still asserts Silent+Programming/Proguntil authenticated final verification clears it. This regression's scope is complete; this parent issue remains open only for its other listed acceptance criteria.[-]Validate final-image buttons, menus, reset stability, and buzzer cues on hardware[/-][+]Validate final-image buttons, menus, reset stability, and audio cues on hardware[/+]atomicdeploy commentedon Aug 9, 2026
Acceptance disposition for the silent delivery pass
This issue owns final-image physical front-panel key perception and menu acceptance, including immediate down/up behavior from #27/#28. Motion with real loads belongs to #72; harmless macro streaming belongs to #74.
Its audio/startup-melody checks cannot be claimed during the user-mandated silent pass. Silence is a safety constraint, not an audio test pass. The current board is not enumerated, the latest verified backup reports
Silent=true, and this reconciliation did not probe, flash, actuate, or play sound. Keep hardware acceptance blocked until David is present and explicitly confirms the safe gate.atomicdeploy commentedon Aug 13, 2026
PR #237 restores and unit-tests the exact autonomous door and motion/output tones. Physical acceptance still requires observing door open/close and loaded motion/output activation/release on the exact flashed artifact; keep this issue open until that hardware evidence is recorded.
atomicdeploy commentedon Aug 13, 2026
Merge-safe P1 reconciliation —
fbb66f215b63476a54ca101f404c6bb50ff5bb32This branch was recovered in the clean worktree
C:\Users\David\AppData\Local\PCController\worktrees\pr271-complete. It contains only normal merge commits—no rebase, force push, or mutation of the evidence checkout. Currentorigin/main9f2782cf0346464eb91a6bc122eeb4c0aa3f9f76is an ancestor of this head.Persisted MENU/MOVE policy
MOVE) remains a directMENU_SET_PAGEcompatibility selector for the unified page-9KEYmotion surface. It is not a separately browsable local page.KEYprogram-mode ID. Host parsing uses that mode to distinguish this alias from the historical genuine MOVE page.Evidence
node --test Tools/Firmware/firmware.test.mjs: 28/28 passed.firmware_controls_tests: passed; it covers persisted legacy mask/default/order migration and RF rank preservation. No VirtualBoard runtime source was changed.git diff --check: passed.Tools/Build/go-tests.mjs; it exceeded the shared product runner's 5-minute tool window without emitting a source failure, so it is not claimed green.controller.exe(the normalgo runis correctly blocked by the workstation's deny-execute Go-temp ACL). It fails at the existing fixed.firmware_identity/.dataoverlap: identity0x7E74..0x7E7F;.data0x7DBC..0x7E87. This is SRAM/identity placement, not a flash overflow; no size claim is made for this head. The compiler emitted only one non-fatal enum conditional warning before the linker failure.No flash, serial, sound, relay, or VirtualBoard runtime action occurred. Keep this PR Draft until the exact-head CI and resource gate are green. Remaining #69 physical acceptance is deliberately untouched.
Related #28 and #69.
atomicdeploy commentedon Aug 13, 2026
CAFE final-image checkpoint for the KEY/motion regression: PR #271 head
cfa47439ea2cff72f33dcf5da7b8ac61d6ace7ff; flashed firmware identity2A866A70(260814005200). Guarded flash/readback/auth passed. Injected K1 lifecycle on the final image passed: Down -> mode 28 / relay mask 0x02; Up -> relay mask 0x00; explicit all-off followed. Firmware now treats KEY as Side A/B motion and implements configured 2-second single-key/opposing-pair/all-four exit with chord immediate stop and release quarantine. Board is returned to KEY page 9, silent=true, relays=0. A 25-second physical-event observation received no key transition (inputs=0xFF, keys=0), so physical hold acceptance remains truthfully open and requires the operator to press the panel. Full all-key/audio matrix remains under this issue; no audio test was performed because night/silent was requested.atomicdeploy commentedon Aug 13, 2026
Follow-up exact-head checkpoint:
87f395ac77a296371bc000c0612fc4cafb3ae86eresolves the non-baseline AVR identity collision without moving the fixed identity region. Fresh source build: 32,250/32,384 flash (134 free), static SRAM 1,459, modeled peak 1,764/2,048 (284 free);npm test --prefix Tools/Firmware29/29 and test-onlyfirmware_controls_testspassed. The earlier repo-health stale-regex failure was fixed in59ca509; exact-head CI is now running. No hardware action. See the PR body for source/app SHA-256 and full policy.atomicdeploy commentedon Aug 13, 2026
PR #271 CI-repair checkpoint: test-only commit
5b427151f2133fe8b9599c2aec7afaf70ee819f2updates stale host catalog expectations to the documented 14-wire-ID protocol: ID 12 remains a direct non-browsable KEY alias and RF remains ID 13. Product-owned Go runner passed 37/37 packages; firmware contract passed 29/29. AVR source is unchanged from the already measured 32,250/32,384 build. Draft remains pending exact-head hosted CI; no hardware action. #271atomicdeploy commentedon Aug 13, 2026
PR #271 is source/CI ready at
5b427151f2133fe8b9599c2aec7afaf70ee819f2and has the persisted MENU/MOVE alias P1 resolved: #271 (comment)#69 remains open for the separate live physical-panel gesture acceptance. No flash, serial, sound, relay, or hardware action was taken in this recovery.
atomicdeploy commentedon Aug 13, 2026
PR #271 main-refresh checkpoint: clean normal merge
d675a41892ce18d784503ec41bdd4477fc4a2e79includesmain@109f4e2; local source gates are green (stable Go 37/37, vet, firmware 29/29, CMake adapter, fixed-identity AVR 32,254/32,384 with 284-byte modeled SRAM margin). Hosted exact-head CI is queued. No serial, flash, sound, relay, or physical-board action occurred.#69 remains the separate post-merge physical-panel acceptance gate. Its existing
🔥 priority: critical,🎛️ front-panel,🧪 testing, and🔍 needs-hardwarelabels remain accurate. Next owner: Integration after #271 exact-head CI and merge. #271atomicdeploy commentedon Aug 13, 2026
PR #271 second main-refresh checkpoint: clean normal merge
dae37d83dd407d6825b83453bd46aa1bc63cdb81includesmain@ce8a837; local source gates passed (stable Go 37/37, vet, firmware 29/29, CMake adapter, fixed-identity AVR 32,254/32,384 with 284-byte modeled SRAM margin). Hosted exact-head CI is running. No serial, flash, sound, relay, or physical-board action occurred.#69 remains the separate post-merge physical-panel acceptance gate. Existing
🔥 priority: critical,🎛️ front-panel,🧪 testing, and🔍 needs-hardwarelabels remain accurate. Next owner: Integration after #271 CI and merge.DRSDavidSoft commentedon Aug 13, 2026
PR #271 is now source/CI complete at dae37d8 (all hosted checks green, current main ce8a837 included). The persisted MENU/MOVE alias is resolved: ID 12 remains direct-only KEY compatibility and RF remains ID 13.
No serial, flash, sound, relay, or physical-board action occurred. #69 therefore remains the distinct post-merge live front-panel acceptance gate. Existing priority-critical, front-panel, testing, and needs-hardware labels remain accurate; no taxonomy gap.
Next owner: Integration may merge #271, then coordinate human-safe physical acceptance here. #271
Restore unified KEY motion and exit gestures (#271)
atomicdeploy commentedon Aug 23, 2026
Hardware acceptance remains required after PR #284: validate specific relay/motion tones without a leading generic beep, global Silent suppression, door open/close cues, physical door notification source/context, and reconnect after a changed USB port assignment. Source/unit/AVR gates are complete; this comment intentionally does not claim live-board acceptance.
Recover edge delivery, USB rebinding, and branded notifications (#284)