Files
wadamesh/tdisplay_p4/main/idf_component.yml
T
Kaj SchittecatandClaude Opus 4.8 2353c56adc p4: T-Display P4 port — web VNC/remote, time-sync hardening, brightness + sound
LilyGo T-Display P4 (ESP32-P4 + factory ESP-AT C6) reaches POC: LoRa mesh, Wi-Fi
companion (TCP:5000), GPS, SD, battery gauge, chat persistence, and now:

- Web VNC/REMOTE/terminal via a first-byte router on the single ESP-AT listener
  (HTTP verbs -> WS server, '<' frames -> phone companion; both live at once).
  C6Server gains a _begun gate so never-listening instances stay inert.
- VNC freeze fixed: bounded one-CIPSEND-chunk drains under the WS client mutex
  (a whole-band blocking write starved the main loop), fast-drop of stalled
  mirror clients, SEND-OK wait 8s->4s.
- Wi-Fi robustness: CWJAP? miss-streak tolerance, CIPSERVER liveness verify +
  re-arm (a re-association silently killed the listener until reboot).
- Time sync: HTTP-Date fallback when SNTP is blocked; ClockFloorRTC gains a
  MAX_PLAUSIBLE_EPOCH guard (all boards) after a garbage-future hardware RTC
  read (2043) latched the ratchet.
- Brightness: RM69A10 DCS 0x51 driven live (CC slider + persisted pref).
- Sound: ES8311 + NS4150B notification chimes (P4Audio, esp-bsp-derived
  register sequence on Wire1), full HAS_UI_SOUND surface + CC Screen/Sound
  sliders; ccVolumeReleaseCb's hardcoded tanBeep -> uiSoundPreview.
- tdisplay_p4/: IDF project (build.sh, c6_at AT-over-SDIO driver, sdkconfigs,
  fetch-deps.sh vendoring, .gitignore for local payloads).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-15 19:13:23 +02:00

25 lines
2.0 KiB
YAML

dependencies:
idf: ">=5.5.1"
# ── Wi-Fi/BLE over the C6 is NOT supported on this board (2026-07-14 investigation) ─────────────
# The T-Display P4's FACTORY C6 runs ESP-AT firmware (SDIO), NOT an esp-hosted slave — exactly what
# Meck-P4's real firmware talks to (Cpp_Bus_Driver::Esp_At, not esp_hosted_connect_to_slave). Our
# whole Wi-Fi/BLE stack is Arduino WiFi over esp-hosted/esp_wifi_remote, so the C6's SDIO stream is
# misframed ("len>max ... Drop") and the P4 reset-loops — this happens at BOTH esp_hosted 2.12.9 and
# 2.0.17 (verified), because it's a firmware-TYPE mismatch, not a version one. Making Wi-Fi work would
# need either (a) re-flashing the C6 with an esp-hosted slave image (the C6 is SDIO-only — no user
# flash path, unshippable) or (b) re-architecting Wi-Fi around the AT interface (large). So the C6 is
# left OFF and TDP4_C6_READY=0; the P4 ships LoRa + UI only. Keep arduino-esp32 at the standard 3.3.10.
badgeteam/badge-bsp: "^0.9.9"
nicolaielectronics/tanmatsu-wifi: "^1.1.2"
nicolaielectronics/tanmatsu-lora: "^0.3.0" # SX1262 LoRa over the C6 — TanmatsuLoraRadio bridge
# NB: tanmatsu-wifi drags in the nicolaielectronics/esp-hosted-tanmatsu FORK (also pulled via
# badge-bsp), whose auto-init constructor grabs the C6 SDIO bus before app_main and blocks our AT
# driver. We can't just remove the dep (it's transitive), so build.sh neutralizes that fork's
# constructor too (alongside espressif/esp_hosted's). The P4's Wi-Fi is AT-over-SDIO via c6_at.
# arduino-esp32 as an IDF component so the shared wadamesh src/ (Arduino API) compiles.
espressif/arduino-esp32: "3.3.10"
# BLE companion over the C6: esp-nimble-cpp is the IDF-host NimBLE C++ wrapper (works with esp-hosted's
# remote BLE controller, unlike NimBLE-Arduino's bundled host). The Tanmatsu's vendored SerialBLEInterface
# copy is ported to this 2.x API. C6 controller comes up via hostedInitBLE().
h2zero/esp-nimble-cpp: "^2.3.0"