GitLab

Fix two "format not a string literal and no format arguments" warnings.

And use a copy-pasted version of geocode-glib's GeocodeLocation
object instead, so we don't end up duplicating bug fixes in
both projects.

So we can get rid of the custom GClueLocationInfo object.

The freegeoip server we use has a warning on it that it can block
clients if they exceeed limits and limits are pretty low.

Our plan has been to switch to our own server usage once its setup but
seems like its goint to happen anytime soon:

https://bugzilla.gnome.org/show_bug.cgi?id=706886

So lets use Fedora's geoip server for now.

... now that this is handled in service-client instead.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

It fits in here more naturally as we are already keeping around the
"old" and "new" location objects whos distances we need to compare.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

https://bugs.freedesktop.org/show_bug.cgi?id=69105

We can pass it directly to the GClueLocationInfo construction now that
it no longer warns about NULL descriptions.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

This reverts commit 931f0133.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

No need to use g_clear_pointer here as g_free already does the right
thing with NULL pointer.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

Document the DistanceThreshold property and LocationUpdated signal.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

set_description with NULL is a valid way for clearing the value.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

https://bugs.freedesktop.org/show_bug.cgi?id=69105

Instead of copying values over from location_info to service_location,
just keep a ref to location_info and let gobject property binding keep
the properties in sync.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

Make sure we don't compare kilometers to meters.

https://bugs.freedesktop.org/show_bug.cgi?id=69105

https://bugs.freedesktop.org/show_bug.cgi?id=69105

Dist org.freedesktop.GeoClue2.service.in so that it gets included in
tarballs.

https://bugs.freedesktop.org/show_bug.cgi?id=69102

https://bugs.freedesktop.org/show_bug.cgi?id=69102

Drop the Requires: as we don't ship a shared library, and add
datarootdir to stop autotools from complaining.

https://bugs.freedesktop.org/show_bug.cgi?id=69102

Ship gnome-compiler-flags.m4 in m4/ so that we wouldn't need the whole
gnome-common during build time.

https://bugs.freedesktop.org/show_bug.cgi?id=69102

Drop the --enable-strict-cc configure option and replace it with
--enable-compile-warnings from gnome-common. This is enabled by default
to make sure warnings show up during development.

https://bugs.freedesktop.org/show_bug.cgi?id=69102

https://bugs.freedesktop.org/show_bug.cgi?id=69102

We don't need to hold a reference to the locator, as this
stops it from getting unreferenced from outside the object,
and thus the async operations from getting cancelled.

https://bugs.freedesktop.org/show_bug.cgi?id=69049

https://bugs.freedesktop.org/show_bug.cgi?id=69049

We were always starting an IP based geolocation, whether or
not the network was available. We should only start it
if the network is available.

https://bugs.freedesktop.org/show_bug.cgi?id=69048

This should have been part of commit 698602ca.

Cancellation of ipclient search request is from us so no need to print
warnings about this error.

So that we are not wrong to destroy associated data immedialy after.

Keep locator alive until gclue_ipclient_search_async() call handling is
done.

Keep GCancellable and IPClient instances alive until the async
soup_session_queue_message() handling is done.