Upstream weekly check, 2026-10-19 to 2026-10-26
Written in our own words. Facts and ids only, each with its source. This is the weekly safety net behind the daily KB run (see UPSTREAM_DAILY_2026-10-26.md); items already handled there are not repeated.
Window
| Source | From | To | Commits | Tuya-relevant |
|---|---|---|---|---|
| Koenkk/zigbee-herdsman-converters | bb5242d9f5 | 7462b541b3 | 45 | tuya.ts, moes.ts, nova_digital.ts, lib/tuya.ts |
| zigpy/zha-device-handlers | 1d6598c32a | 1f1405b0eb | 3 | tuya/mcu 0x25 reply framing (already checked by the daily run) |
| dresden-elektronik/deconz-rest-plugin | a4c17adfa0 | same | 0 | none |
| blakadder/zigbee | d96e8aaf72 | same | 0 | none |
| make-all/tuya-local | 225943071b | 8de6b3cce6 | 35 | Wi-Fi devices only, no Zigbee couple |
| JohanBendz/com.tuya.zigbee | 130854f4c6 | 61577d8ec2 | 1 | workflow only |
Implemented this run
- P3060: packed phase DPs (8-byte voltage/current/power) can carry negative power in two ways. The offset form (raw minus 0x99999A) was already decoded; some
_TZE200_nslr42ttunits use plain 24-bit two's complement above 0x99999A instead, which we showed as a power of several MW. Both are now decoded inlib/tuya/TuyaNegPower24.js, used byTuyaPhaseStructandSpmPhaseDecoder. Values below 0x99999B decode exactly as before. Source: zhc 7462b54src/lib/tuya.ts(Koen Kanters and contributors). Test:test/p3060-neg-power-24.test.js.
Parked couples (publish budget still full, same reason as the daily run)
| Couple | Device (source) | Target driver | Note |
|---|---|---|---|
_TZE204_kobbcyum / TS0601 | Tongou TOWSMR1-40A-AC RCBO family (zhc 7462b54 tuya.ts, joins _TZE284_432zhuwe) | smart_rcbo | today only in climate_sensor, see leads |
_TZE2841000000_8se38w3c / TS0601 | long-name form of TZ-ZT01_GA4 temp/humidity probe (zhc 7462b54 tuya.ts) | climate_sensor | Homey matches the exact string, so the long form needs its own entry |
_TZ3210_cmnwaylf / TS0501B | ECODO ECD-ZB-REPEATER, a repeater with an indicator LED driven as a light (zhc 7462b54 tuya.ts) | zigbee_repeater (LED as optional onoff/dim) | absent everywhere |
Wrong-class leads (decision needed, not moved)
_TZE284_tdhnhhiy/ TS0601: Nova Digital SYZB-6W, a 6-gang wall switch (DP1-6 gangs, DP16 backlight). Z2M moved it out of the generic 6-gang list into its own definition innova_digital.tsthis week. We list it in climate_sensor only. Target: wall_switch_6_gang_tuya._TZE204_nvxorhcj/ TS0601: Nova Digital ZTS-8W-B, 8 gangs (DP1-6 plus DP101/102). We list it in switch_1gang._TZE284_kow4ok3t/ TS0601: Nova Digital ZTS-8B, 8 gangs, same layout. We list it in switch_4_gang_metering. There is no 8-gang driver yet; per the standing rule a newwall_switch_8_gang_tuyacategory is the clean fix, to ship as its own isolated master step after headroom lands._TZE204_kobbcyumand_TZE284_432zhuwe/ TS0601: Tongou RCBO, listed in climate_sensor. Target smart_rcbo._TZE200_nslr42tt/ TS0601: bidirectional 3-phase meter in Z2M, listed in switch_wireless (and handled in power_clamp_meter code without a manifest claim). Target power_clamp_meter or energy_meter_3phase after a DP check.
Upstream changes of substance
- Z2M split three Nova Digital multi-gang switches out of the generic Tuya lists into dedicated definitions (new file
nova_digital.ts), and split_TZE204_6a4vxfnvaway from theedl8pz1kfloor thermostat into a new definition shared with_TZE204_0hhou4yj(ELECTSMART). The floor thermostat part is already in the daily leads. - More long manufacturer names (
_TZE28C1000000_*,_TZE2841000000_*) appeared as siblings of existing couples. Our runtime already treats that prefix as EF00 (ManufacturerVariationManager), but each long name still needs its own manifest entry to pair. _TZE284_hkdl5fmv(new Z2M sibling of the_TZE200_hkdl5fmvmeter) is already in energy_meter_din on master.- No couple was removed upstream this week except through the moves above.