skruecken/powerpods
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add CACHE_STATUS UART poll and dashboard live stream.

Browse Source 
Combine cached accel and tap in one low-overhead master command for ~16 ms
host polling. The dashboard uses a single live-stream toggle plus per-slave
accel-stream controls; fix live_stream state so polling is not cleared every
slow client refresh.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
skruecken
2026-05-29 20:57:54 +02:00
co-authored by Cursor
parent a8d4d42920
commit f512936d97
21 changed files with 764 additions and 377 deletions
Download Patch File Download Diff File Expand all files Collapse all files
main/README.md
+13
View File
@@ -225,6 +225,7 @@ Host and master speak nanopb-encoded `UartMessage` inside UART frames (byte 0 =
| 24 | `ACCEL_SNAPSHOT` | Implemented (`cmd/cmd_accel_snapshot.c`) — cached slave accel from ESP-NOW stream |
| 27 | `TAP_NOTIFY` | Implemented (`cmd/cmd_tap_notify.c`) — get/set which tap kinds notify via ESP-NOW |
| 28 | `TAP_SNAPSHOT` | Implemented (`cmd/cmd_tap_snapshot.c`) — consume cached tap events from master registry |
| 29 | `CACHE_STATUS` | Implemented (`cmd/cmd_cache_status.c`) — combined accel + tap cache in one UART round-trip |
Regenerate C code:
@@ -367,6 +368,17 @@ go run . -port /dev/ttyUSB0 tap-notify -client 16
Read **cached** tap events on the **master** (one pending event per slave). Slaves send `ESPNOW_TAP_EVENT` on tap; the master stores the latest value per client in `client_registry.c` for up to **16 ms** (`CLIENT_REGISTRY_TAP_MAX_AGE_MS`). Each snapshot **consumes** fresh events (cleared after read).
### CACHE_STATUS command
Fast combined poll for host tools at **16 ms** or faster: one UART frame, no request body (command id `0x1d` only).
**Response:** `cache_status_response` with:
- `accel[]` — same as `ACCEL_SNAPSHOT`, only clients with `accel_stream_enabled`
- `taps[]` — same as `TAP_SNAPSHOT`, only clients with any tap notify flag; pending taps are consumed
The master walks `client_registry` once per request (`cmd/cmd_cache_status.c`). Prefer this over separate `ACCEL_SNAPSHOT` + `TAP_SNAPSHOT` when polling both streams.
Only slaves with at least one tap-notify flag enabled are included.
**Request:** framed `1c` (`0x1c`) + optional `tap_snapshot_request` (`client_id`: `0` = all, `>0` = one id).
@@ -555,6 +567,7 @@ Target: ESP32-S3. Close serial monitor on the UART adapter port before running `
| `cmd/cmd_accel_snapshot.c` | UART `ACCEL_SNAPSHOT` — cached slave accel |
| `cmd/cmd_tap_notify.c` | UART `TAP_NOTIFY` — ESP-NOW tap notify config |
| `cmd/cmd_tap_snapshot.c` | UART `TAP_SNAPSHOT` — consume cached tap events |
| `cmd/cmd_cache_status.c` | UART `CACHE_STATUS` — combined accel + tap cache poll |
| `board_input.c/h` | Taster GPIO12, LiPo ADC on GPIO1 / GPIO12 |
| `pod_settings.c/h` | NVS persistence (accel deadzone, …) |
| `led_ring.c/h` | LED ring (digit display, progress bar) |