[WIP] Gweather

[yetist]

Use libgweather instead libmateweather.

[rbuj]

The locations are not displayed correctly after adding or removing them from the Edit dialog, showing duplicated locations.

[rbuj]

I got this log:

$ journalctl --no-pager --boot
<cut>
Jun 01 13:55:00 rawhide clock-applet[42142]: invalid unclassed pointer in cast to 'ClockLocationTile'
Jun 01 13:55:00 rawhide clock-applet[42142]: clock_location_tile_refresh: assertion 'CLOCK_IS_LOCATION_TILE (this)' failed
Jun 01 13:55:00 rawhide clock-applet[42142]: invalid unclassed pointer in cast to 'ClockLocationTile'
Jun 01 13:55:00 rawhide clock-applet[42142]: clock_location_tile_refresh: assertion 'CLOCK_IS_LOCATION_TILE (this)' failed
Jun 01 13:55:00 rawhide clock-applet[42142]: clock_location_tile_get_location: assertion 'CLOCK_IS_LOCATION_TILE (this)' failed
Jun 01 13:55:00 rawhide clock-applet[42142]: clock_location_compare: assertion 'CLOCK_IS_LOCATION(one)' failed
Jun 01 13:55:00 rawhide clock-applet[42142]: invalid unclassed pointer in cast to 'ClockLocationTile'
Jun 01 13:55:00 rawhide clock-applet[42142]: clock_location_tile_refresh: assertion 'CLOCK_IS_LOCATION_TILE (this)' failed
<cut>

[raveit65]

My results are different, in fedora 32 and applets are build-in-process.
I don't have double entries, but i have to close/reopen the clock window after adding the locations before i see them.
But i don't have any temperature values.
Logs seems to be OK here.

[rave@f32 ~]$ journalctl -b0 -r --user --grep=clock-applet
-- Logs begin at Wed 2020-02-12 00:48:49 CET, end at Mon 2020-06-01 20:58:20 CEST. --
-- No entries
[rave@f32 ~]$ su
Password: 
[root@f32 rave]# journalctl -b0 -r --grep=clock-applet
-- Logs begin at Wed 2020-02-12 00:47:41 CET, end at Mon 2020-06-01 20:56:24 CEST. --
-- No entries --
[root@f32 rave]# 

[lukefromdc]

I just tested this (with an in-process build as I usually use). Builds fine and nothing crashes. As of 6-11-2020, adding a location works after a little fiddling (but does not crash), and weather can then be shown though not temperature. I am assuming the latter issue is the same server issue discussed at mate-desktop/libmateweather#79

[N0rbert]

I have to share my concern which I have wrote on mate-desktop/libmateweather#79 :

Personally I do not like GNOME parts to be included to MATE. This will lead to lose of traditional desktop. We can't predict what GNOME developers will remove next time. Dependance on non-stable GNOME' developer minds is not good idea.
Also I do not think we need Vala on MATE, as GWeather depends on Vala.
So I think that defining correct aviationweather URL in the dconf/gsettings (while using libmateweather) is better solution.

This is what I think, so staying with libmateweather and leaving things as they were is better and safer.

[raveit65]

@N0rbert
Please test things before your are posting again wrong facts.

You don't need to install vala for using libgweather. This is a developer language and not needed for running libgweather.
By the way, switching to libgweather is on our roadmap for a long time.

Again, we are a small team an we have a critical work overload.
So please stop saying what we have to do with our spare time.
And please blame your ubuntu maintainer!!! He did not answering on posts, notifications or working on code here since more than 2-3 years.
This is one reason why other team members have a work overload.
Thank you.

[N0rbert]

@raveit65
I agree about Vala, sorry about this. It seems you forgot that I'm Ubuntu MATE team member, so I will not continue this discussion. I do not blame anyone. Just wanted to help. Did you forget who contacted with NOAA officials? I'm doing this in my spare time for the love of both MATE and Ubuntu MATE which are both last resort for normal adequate desktop computing in 2020. So hold on and chill out.

[raveit65]

Simply test your opinion before you post them :)
And switching to libgweather will reduce our work load.

[ssrublev]

MATE and Ubuntu MATE which are both last resort for normal adequate desktop computing in 2020.

Ubuntu MATE, sadly, is loosing "last resort" sense because of loosing basical mate-panel "systray" functionality and not willing to fix it... I mean https://ubuntu-mate.community/t/i-cannot-disable-sni-in-mate-panel-by-setting-enable-sni-support-to-false/17187/7

[lukefromdc]

Ubuntu has by default (last I recall) the separate indicator applet in the panel, in fact it was Ubuntu that first coded this I'm pretty sure. To force everything to status notifier should work if you remove the indicator applet (s) from the panel as well as making sure SNI support is set to false. If you do this and anything still shows up as an indicator (in SNI mode) that is a bug.

[raveit65]

a) I can confirm now the issue with seeing double locations when adding or removing one location.
b) The UI (CalendarWindow) is jumping/starts resizing after opening, caused by adding the weather icons to ClockLocationTile.
c) The UI (CalendarWindow) is jumping/starts resizing when activating the edit-button with mouse hover in ClockLocationTile, especially the locations have long descriptions (eg. Kaumalapau, USA, Hawai).

The result of jumping/starts resizing is that the CalendarWindow moves out of the monitor, you can see the missing edge on a secondary monitor.

PS: Weather data is working now again.

[lukefromdc]

Any updates on this? I would prefer to keep this open as finishing this would allow offloading the current libmateweather maintainance, at the price that we would have to bring it back if anything bad ever happened to libgweather. Note that to DROP libmateweather would also require porting the panel applet to libgweather