v0.1.0a4 -- part B of the survey, and a factor-10 error found and fixed
The control test writes each control, checks it on three legs and puts it
back. On its first daylight run it found that register 13074, the export power
limit, reads in watts and writes in tens of watts -- so both this integration
and the YAML package were setting ten times what the user asked for. Measured
six times at one house and replicated at two more, on three models over both
transports; fixed here, still open in the YAML package.
Required rather than optional: the integration imports
sungrow_modbus.control_test and reads field_names, probed_capabilities
and inverter_serial off library objects, none of which 0.1.0a3 has.
Annotated, because git push --follow-tags ignores lightweight tags -- which
is how v0.1.0a1 reached nothing at all.