Hi, There are several bug reports about redshift not being able to get the location from geoclue: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=789883 https://bugzilla.redhat.com/show_bug.cgi?id=1214978 https://github.com/jonls/redshift/issues/158 Reading https://bugzilla.redhat.com/show_bug.cgi?id=1214978#c3 it shouldn't be necessary to change the configuration file, but without explicit authorization it's definitely not working.
(In reply to Laurent Bigonville from comment #0) > Hi, > > There are several bug reports about redshift not being able to get the > location from geoclue: > > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=789883 > https://bugzilla.redhat.com/show_bug.cgi?id=1214978 > https://github.com/jonls/redshift/issues/158 > > Reading https://bugzilla.redhat.com/show_bug.cgi?id=1214978#c3 it shouldn't > be necessary to change the configuration file, but without explicit > authorization it's definitely not working. As I commented on the RHEL bug, it's either a bug in the app or geoclue itself but redshift doesn't need any special config in geoclue.
I think this is a user error (see comment on redhat bug). If it turns out that I'm wrong, please do re-open this bug.
Well I also have the same issue. The geolocation parameter in the privacy tab is on
$ G_MESSAGES_DEBUG=all /usr/lib/geoclue-2.0/geoclue (geoclue:23157): Geoclue-DEBUG: Available accuracy level from GClueWifi: 4 (geoclue:23157): Geoclue-DEBUG: New agent for user ID '1000' (geoclue:23157): Geoclue-DEBUG: WiFi device 'wlan0' added. (geoclue:23157): Geoclue-DEBUG: Available accuracy level from GClueWifi: 6 (geoclue:23157): Geoclue-DEBUG: Number of connected clients: 1 (geoclue:23157): Geoclue-DEBUG: New distance threshold: 50000 (geoclue:23157): Geoclue-DEBUG: 'redshift' not in configuration (geoclue:23157): Geoclue-DEBUG: requested accuracy level: 4. Accuracy level allowed by agent: 8
(In reply to Laurent Bigonville from comment #4) > $ G_MESSAGES_DEBUG=all /usr/lib/geoclue-2.0/geoclue > (geoclue:23157): Geoclue-DEBUG: Available accuracy level from GClueWifi: 4 > (geoclue:23157): Geoclue-DEBUG: New agent for user ID '1000' > (geoclue:23157): Geoclue-DEBUG: WiFi device 'wlan0' added. > (geoclue:23157): Geoclue-DEBUG: Available accuracy level from GClueWifi: 6 > (geoclue:23157): Geoclue-DEBUG: Number of connected clients: 1 > (geoclue:23157): Geoclue-DEBUG: New distance threshold: 50000 > (geoclue:23157): Geoclue-DEBUG: 'redshift' not in configuration > (geoclue:23157): Geoclue-DEBUG: requested accuracy level: 4. Accuracy level > allowed by agent: 8 There were more questions on the RH Bugzilla bug: - does it have a .desktop file - please link to the code that registers the .desktop file with Geoclue
> There were more questions on the RH Bugzilla bug: > - does it have a .desktop file Alright, so indeed adding a .desktop file is fixing the issue.. I'm not sure what you mean by: > - please link to the code that registers the .desktop file with Geoclue
(In reply to Laurent Bigonville from comment #6) > > There were more questions on the RH Bugzilla bug: > > - does it have a .desktop file > > Alright, so indeed adding a .desktop file is fixing the issue.. > > I'm not sure what you mean by: > > > - please link to the code that registers the .desktop file with Geoclue A link to the code in Redshift that uses Geoclue2 would have been needed, if indeed it had a .desktop file.
Oh https://github.com/jonls/redshift/blob/master/src/location-geoclue2.c#L202
(In reply to Laurent Bigonville from comment #6) > > There were more questions on the RH Bugzilla bug: > > - does it have a .desktop file > > Alright, so indeed adding a .desktop file is fixing the issue.. Cool. Problem solved then?
(In reply to Laurent Bigonville from comment #8) > Oh > > https://github.com/jonls/redshift/blob/master/src/location-geoclue2.c#L202 That's not a valid .desktop ID. Geoclue should probably warn loudly when the .desktop ID is invalid, or not present on the system. Reopening to that effect.
(In reply to Bastien Nocera from comment #10) > (In reply to Laurent Bigonville from comment #8) > > Oh > > > > https://github.com/jonls/redshift/blob/master/src/location-geoclue2.c#L202 > > That's not a valid .desktop ID. Geoclue should probably warn loudly when the > .desktop ID is invalid, or not present on the system. Reopening to that > effect. It's not geoclue that's checking it, it's the agent (gnome-shell). This is more compatible with how things will be done in post-sandboxing world, I'm told.
(In reply to Zeeshan Ali from comment #11) > (In reply to Bastien Nocera from comment #10) > > (In reply to Laurent Bigonville from comment #8) > > > Oh > > > > > > https://github.com/jonls/redshift/blob/master/src/location-geoclue2.c#L202 > > > > That's not a valid .desktop ID. Geoclue should probably warn loudly when the > > .desktop ID is invalid, or not present on the system. Reopening to that > > effect. > > It's not geoclue that's checking it, it's the agent (gnome-shell). This is > more compatible with how things will be done in post-sandboxing world, I'm > told. Then we should make sure that that error is set in the agent, and have geoclue print the error as well, before passing the rejection back to the application.
(In reply to Bastien Nocera from comment #12) > (In reply to Zeeshan Ali from comment #11) > > (In reply to Bastien Nocera from comment #10) > > > (In reply to Laurent Bigonville from comment #8) > > > > Oh > > > > > > > > https://github.com/jonls/redshift/blob/master/src/location-geoclue2.c#L202 > > > > > > That's not a valid .desktop ID. Geoclue should probably warn loudly when the > > > .desktop ID is invalid, or not present on the system. Reopening to that > > > effect. > > > > It's not geoclue that's checking it, it's the agent (gnome-shell). This is > > more compatible with how things will be done in post-sandboxing world, I'm > > told. > > Then we should make sure that that error is set in the agent, and have > geoclue print the error as well, before passing the rejection back to the > application. That will require changes in agent API and hence incompatibility between newer geoclue and existing gnome-shell. :( Maybe upon rejection from agent, geoclue could just emit a slightly more helpful message: "Agent rejected '%s' for user '%u'. Please ensure that '%s' has installed a valid %s.desktop file."?
(In reply to Zeeshan Ali from comment #13) > (In reply to Bastien Nocera from comment #12) > > (In reply to Zeeshan Ali from comment #11) > > > (In reply to Bastien Nocera from comment #10) > > > > (In reply to Laurent Bigonville from comment #8) > > > > > Oh > > > > > > > > > > https://github.com/jonls/redshift/blob/master/src/location-geoclue2.c#L202 > > > > > > > > That's not a valid .desktop ID. Geoclue should probably warn loudly when the > > > > .desktop ID is invalid, or not present on the system. Reopening to that > > > > effect. > > > > > > It's not geoclue that's checking it, it's the agent (gnome-shell). This is > > > more compatible with how things will be done in post-sandboxing world, I'm > > > told. > > > > Then we should make sure that that error is set in the agent, and have > > geoclue print the error as well, before passing the rejection back to the > > application. > > That will require changes in agent API and hence incompatibility between > newer geoclue and existing gnome-shell. :( Ha, you should probably have used D-Bus errors when rejecting clients, instead of simply sending true/false. > Maybe upon rejection from agent, > geoclue could just emit a slightly more helpful message: "Agent rejected > '%s' for user '%u'. Please ensure that '%s' has installed a valid %s.desktop > file."? That would be good enough for me, thanks.
commit: ca7fb0ced37cd6b57ef7693eb05d99b4646b8abb service-client: More helpful error on agent rejection https://bugs.freedesktop.org/show_bug.cgi?id=91237
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.