I'm still testing the skin out, so not sure I can recommend it to others full-time. But if you're interested, it's available from https://bitbucket.org/pjhardy/weewx-opensprinkler
So far I've only implemented the Zimmerman Method, partly because I'm not sure how to get forecast data in to weewx either. :) In my install I'm using the Responsive skin as well, which includes an extension for getting forecasts from the Australian BoM. But that's hardly a general solution.
I’ve been messing around with this for a few months. Initially, the external weather offset
was always returning a scale of zero because of how it was pulling OWM data, so I brought that
service inside the house and updated the OWM code to work better. I also implemented the
same thing using DarkSky data and calculate the scale from both. The external thing runs on
NodeJS, which was pain to get running on an RPi, so I’m rewriting the whole mess in python.
What’s left is to get it working as a CGI server so the OS controller can talk to it. In terms of weewx
data, all I’m pulling at the moment is the last ~2 days of rain using a template to write a web
page the new code fetches. That gets added to the forecast calculations. The scale numbers
between old way and new code track, but often don’t match because of the math related to the
3 parameters (temp and humidity mostly), which vary slightly between OWM and DS.
No idea if the OS guys every fixed the forecasting thing or not, but they knew it was a problem.
I’d rather have all this inside my network anyway. The WU thing only works when you have a key,
and WU quite giving them away, which is why the OS guys switched to the OWM api.
try the weewx-forecast extension. it can pull forecasts from a few different sources, and it sticks the results into a dedicated forecast database. then you can query that database from your service to get the specific rainfall information you want.
There could also be some value in trying to bring BoM support in to weewx-forecast. I'm not going to commit to that until I've actually done it though. ;)
So, I was able to port the OpenSprinkler weather server to python as a cgi script. It isn't ready for public consumption, but does cover most of the capabilities of the original. It queries Open Weather Map and Dark Sky using my private keys for each, as well as pulling in a weewx generated web page with the last 2 days rain, so that part would have to be figured out. It does require a local web server at this point. I wasn't able to (didn't try too hard) to get a python web server to simply serve the cgi results internally verses running the same code via cgi mechanisms. The scale from the two sources are averaged for now. If anyone is interested I can package in some form or another.
Sounds interesting! I use OpenSprinkler but never used the watering calculations - have always wanted to look into it but never have yet.
Il giorno martedì 2 ottobre 2018 23:41:40 UTC+2, Pat ha scritto:Sounds interesting! I use OpenSprinkler but never used the watering calculations - have always wanted to look into it but never have yet.Do you take into account the Evapotranspiration value ?