No idea what is actually sane. Nobody is hopefully running this on lower
than VGA resolution.

Dropped in 060a9a2a.

Add a combo box with text entry that accepts a MAC address and also
the selection of one of the predefined special values.

It is not possible to configure properties like the generated MAC
address mask or the stable id. For these advanced settings users
should rely on nmcli.

NetworkManager supports it now.

Connections that become invisible initially emit a CHANGED signal when
their settings are removed and then they will disappear.

Ignore invisible connections and ones that are still visible but don't
have a NMSettingConnection since they are being removed.

https://bugzilla.redhat.com/show_bug.cgi?id=1410742

Only attempt to create an shared application instance only if there actuall is
a bus that would allow its reuse.

This seems like a good idea. Apart from a tiny speed improvement by
avoiding directory lookups when loading the UI elements and condensing
the XML data by whitespace ommission, this makes it easy to run the
applet and the connection editor without installing them.

This makes it possible to run the binaries directly from the build tree
and not pick up the glade files from whatever happens to be installed in
the system. Neat!

Elements of the list returned by gtk_size_group_get_widgets() become
invalid after the call to gtk_size_group_remove_widget().

https://bugzilla.redhat.com/show_bug.cgi?id=1380424

Hide the device MAC address from the device combo when the device
doesn't have a permanent MAC.

https://bugzilla.redhat.com/show_bug.cgi?id=1380424

It's always TRUE.

https://bugzilla.gnome.org/show_bug.cgi?id=750261

gtk_tree_model_get() copies the returned boxed value, i.e. it calls
wireless_security_ref() and thus the element must be unreferenced
afterwards.

This causes memory leaks as the WirelessSecurity is never freed due to
circular references: WirelessSecurity references the EAPMethods
through the combo box model, and EAPMethods reference the
WirelessSecurity back.

EAP methods are instantiated by the WirelessSecurity and are supposed
to have a lifetime not longer than the WirelessSecurity's (as they are
hooked in the combo box model and destroyed when the WS is disposed),
so the reference is not needed.

https://bugzilla.gnome.org/show_bug.cgi?id=776384

Previously we only checked the existence of the proxy; we should
instead verify that MM is actually running.

Broadband devices stay uninitialized if they are added before the
ModemManager is running. Call again the device_added method after MM
becomes available.

https://bugzilla.gnome.org/show_bug.cgi?id=776384

Jonathan Kang authored Nov 24, 2016 and Beniamino Galvani committed Dec 21, 2016

Currently, nm-applet only support TTLS/EAP-MSCHAPV2. It makes users
unable to connect to wifis that only supports TTLS/MSCHAPv2.

https://bugzilla.gnome.org/show_bug.cgi?id=768489

Grammar, capitalization, trailing spaces, Unicode typography.

https://bugzilla.gnome.org/show_bug.cgi?id=621767

Atul Anand authored Dec 06, 2016 and Thomas Haller committed Dec 16, 2016

This patch removes multiple entries/widgets following conventional
GUI and introduces basic features following what nmcli provides for
proxy handling and manipulation. It is an unfair idea to follow the co-
ventional GUI and provide entries for http/https/ftp/socks which are
actually browser protocols and standard browsers have already been
provinding proxy feature. On the other side we can't provide entries
for all schemes (there are a lot, smtp://, mail://, gopher://, etc).

Even if nm-c-e provides them, it would only be able to create protocol
based scripts not general ones because the later one require more pa-
rameters than just proxy:port and excludes.

NetworkManager is intended to provide systemwide proxy settings, so it
should look for features that are common for all.