This removes an extra g_object_ref when inserting items to GtkListStore:
the list store takes care of incrementing the reference count itself and
the caller doesn't need to do it.
https://bugzilla.gnome.org/show_bug.cgi?id=709453
Commit b9e3603ba4 added an unref to fix a
memory leak, but it went to a wrong place -- in XML loader, all items
are stored in a hash table that takes ownership of them, and destroyed
when the hash table goes away.
This commit fixes the reference counting in the XML loader and adds
explicit g_object_ref / g_strdup when inserting values to the hash table
to make the memory management more obvious.
The following commit fixes the real leak.
https://bugzilla.gnome.org/show_bug.cgi?id=709453
This caused criticals when on_permission_changed callback tried to
access priv->datetime_dialog or priv->timezone_dialog before they had
been set up.
Another attempt at fixing the leak in the following patch.
This reverts commit 545c886543.
https://bugzilla.gnome.org/show_bug.cgi?id=709443
Instead of setting the size on each child, set it on the parent grid
because only one of those children are visible at any given point.
We are cheating a bit by ignoring the fact there is a second child
which is always shown -- the button box. This will make the other
children lose some height and the dialog will be smaller. However it is
still too big for smaller screens.
https://bugzilla.gnome.org/show_bug.cgi?id=708943
We want to track the busy state of the source. To do that we will keep
a count of the number of pending operations. Not having a check for
NOT_FOUND makes it slightly easier to do that. Not a big deal, but I
think it outweighs the downfall of showing an extra warning.
Plus we are trying to create the directories before using them, so the
warning means something fishy has been going on with your system.
https://bugzilla.gnome.org/show_bug.cgi?id=708943
We receive ClutterEvents, not GdkEvents for button presses on
ClutterActors, so use the correct functions to filter devices.
This also fixes the offset used to access the coordinates of the
events. We were actually using the pointer to the source and x
struct members instead of x and y.
https://bugzilla.gnome.org/show_bug.cgi?id=707784
When enabling DLNA, also make sure to enable the plugin that
we use to export the media to the TV or set-top-box, otherwise
rygel won't export anything.
https://bugzilla.gnome.org/show_bug.cgi?id=708055
Doesn't make sense to keep it sensitive after the dialog returns OK.
This also avoids using finalized objects if the forget button is
clicked after some previously checked rows were destroyed because
their connection got removed elsewhere.
https://bugzilla.gnome.org/show_bug.cgi?id=709118
The check button isn't a direct child of the GtkListBoxRow instance
that contains the "connection" object data so we are adding the wrong
widget to the "rows" list which then gives us a NULL for "connection"
in really_forget() and making us fail to remove the connection.
https://bugzilla.gnome.org/show_bug.cgi?id=709091
