diff --git a/CLAUDE.md b/CLAUDE.md index 7fdee28..d71377e 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -437,6 +437,17 @@ without a fault (`DIAS_OPERACION`); a unit belongs to the site of its most recen the vehicle number, type and 12/24 V. Home per-site counts read the same endpoint (old count until the restart); site goal reads «1–2 unidades instaladas y probadas». Lote `2026-10-22_instalaciones` points each site's install task at both actas. Crisis lote applied by Leo at 14:00; this deploy again needs `restart actas-hbox` before `tablero.sh aplicar`. +**Installation meeting (2026-10-22, 14:00; agreements confirmed by Leo at 16:13).** Gemini recorded it in English, so +what was said in Spanish was lost; Spanish notes in Leo's Drive (Google Meet → the meeting's folder) and the Juntas page now +says to set the meeting language to Spanish before starting. Agreed: field owners ICAVE **Alán**, EIT **Miguel**, LCT +**Antonio**, TIMSA **Jesús Honorato** (field technician and lead in Manzanillo); installation leads **Alán and Daniel**. Two Jesús: the +site user `jesus` is **Jesús Romero**, the installation technician (his actas are signed that way); Miguel, Antonio and +Honorato have no site account yet (`t-alta-responsables-campo`). **D-22-1 decided: GO to the 5 + 5 pilot, but the order waits until the +board's quick wins are done** (`t-v56-piloto` blocked by them; the pilot's order and bench hitos are now dateless). Daily +15-min standup + weekly quick-wins / traffic-light review. LCT is handled separately: the material should be there; only +the ironwork of the RTG brackets is missing (`t-lct-herreria`). Gibran opened the Wialon retranslator task during the meeting +(`T261022-5b37`, Gibran + Jacobo, 26-oct); `t-wialon-retranslator` closed as its duplicate. Lote +`2026-10-22_junta-instalacion`; `proyecto.json` sites carry `responsable`, shown on the home page. **FOTA over 4G — feasibility per board (2026-10-22, 16:30; Leo: «what's needed, is it feasible on every board?»).** Yes on **V5.6/V5.7 with no copper change**: downlink = `PIN_RTCM_TAP` PC7 (USART6_RX ← R111 ← DOUT_4G, 115200), so Helgen's caster can interleave proprietary RTCM 3 frames (4001–4095) carrying signed image chunks that the F9P drops and the MCU collects; diff --git a/hbox-docs b/hbox-docs index 0e49947..e1681a8 160000 --- a/hbox-docs +++ b/hbox-docs @@ -1 +1 @@ -Subproject commit 0e499476fd7dbf8c26e1e9433b2a850a5b88cc9f +Subproject commit e1681a869306b79349e1ad02d078687b0f90aafa