2021-11-08, 23:30
OzWeather on Kodi versions *less than* 19 is currently broken due to very substantial upstream site changes.
OzWeather for Kodi 19+ is not affected - other than if you haven't yet re-configured your addon to use BOM locations (in which case the fallback to the old Weatherzone scraper will now be broken).
For Kodi 19 uses, if you're hitting that, the fix is easy - just go in to OzWeather settings and get your BOM locations setup and it will all work again.
Just to add more context:
OzWeather v1.x is the Kodi (up to 18), Python 2 version - Uses Weatherzone locations, & scrapes Weatherzone pages for its data (Weatherzone in turn got most of the data from the BOM). This worked with minor fixes for 10+ years. But Weatherzone has begun to use less BOM data, and have this last week or so published their entirely new website.
OzWeather v2.x is the Kodi 19+, Python 3 version - it uses an (unofficial) BOM API, getting BOM locations and data directly as JSON. It's a complete re-write & considerably simpler/more stable. I also built the skin patcher tool, and through that added support for the radar features to a bunch of popular skins. So far this has proved to be faster, more robust and is in all quite an improvement. As long as they don't shut off the API (they could of course!) - it should carry on working well.
(Both versions use the BOM FTP side for the radar).
To fix the Kodi 18 version means a re-write of a large section of that addon, in old Python 2, and again in a way that uses scraping (which is much more prone to breakage than using an API of course). I don't know how many folks are still on 18, and I am not sure if/when I will have time to do this. I honestly can't see any good reason, at this point, not to move to Kodi 19, and I am inclined to just draw a line under Kodi 18/Python 2 and move on. But happy to hear others thoughts on this.
(Of course I'd be very happy to accept a PR from someone if they care to fix it & submit such - it's a self contained Python 2 unit, perfect sort of thing for developing and using fairly simple Python skills as a learning exercise...)
OzWeather for Kodi 19+ is not affected - other than if you haven't yet re-configured your addon to use BOM locations (in which case the fallback to the old Weatherzone scraper will now be broken).
For Kodi 19 uses, if you're hitting that, the fix is easy - just go in to OzWeather settings and get your BOM locations setup and it will all work again.
Just to add more context:
OzWeather v1.x is the Kodi (up to 18), Python 2 version - Uses Weatherzone locations, & scrapes Weatherzone pages for its data (Weatherzone in turn got most of the data from the BOM). This worked with minor fixes for 10+ years. But Weatherzone has begun to use less BOM data, and have this last week or so published their entirely new website.
OzWeather v2.x is the Kodi 19+, Python 3 version - it uses an (unofficial) BOM API, getting BOM locations and data directly as JSON. It's a complete re-write & considerably simpler/more stable. I also built the skin patcher tool, and through that added support for the radar features to a bunch of popular skins. So far this has proved to be faster, more robust and is in all quite an improvement. As long as they don't shut off the API (they could of course!) - it should carry on working well.
(Both versions use the BOM FTP side for the radar).
To fix the Kodi 18 version means a re-write of a large section of that addon, in old Python 2, and again in a way that uses scraping (which is much more prone to breakage than using an API of course). I don't know how many folks are still on 18, and I am not sure if/when I will have time to do this. I honestly can't see any good reason, at this point, not to move to Kodi 19, and I am inclined to just draw a line under Kodi 18/Python 2 and move on. But happy to hear others thoughts on this.
(Of course I'd be very happy to accept a PR from someone if they care to fix it & submit such - it's a self contained Python 2 unit, perfect sort of thing for developing and using fairly simple Python skills as a learning exercise...)
