Comment 7 for bug 1377612

Revision history for this message
In , Harald Judt (hjudt) wrote :

Created attachment 5509
make-plugin-ready-for-met.no-locationforecast-1.2-API.patch

I have not received any replies from met.no yet, and though not much time has passed, I'm not very convinced I will. Nevertheless, I have created patches to deal with this problem, it would be great if someone could try them and verify that the results look sane.

The first one prepares the plugin for the new API 1.2, but still uses the old version 1.1. After applying this patch, you should see no differences to an unpatched plugin.

The second patch then switches to the new API 1.2. If you do an immediate comparison, you will probably notice differences in the forecasts, because the new API seems to produce different results than 1.1. If not, you will at least find that the WARNING no longer appears.

While I have tried to match the symbols as good as possible and these patches should keep the 0.8.3 plugin going for some while longer, I have decided to switch to openweathermap for the next feature release. Openweathermap has many benefits, many other clients use it and it has good documentation, especially in comparison to met.no.

At the moment, it is not advised to compile and use current git because there are problems with the update code, noticeable in sunrise/sunset values after 24 hours usage, though there should be no other issues and no crashes. I'm not sure if I will ever fix this, since I'll have to throw away code for openweathermap. Maybe I'll release a stable 0.8.4 (basically 0.8.3 with bug fixes but without new features) if I get to it before working on the provider switch, but I can't promise that.