nixpkgs/nixos/tests/ayatana-indicators.nix

112 lines
3.8 KiB
Nix

import ./make-test-python.nix ({ pkgs, lib, ... }: let
user = "alice";
in {
name = "ayatana-indicators";
meta = {
maintainers = lib.teams.lomiri.members;
};
nodes.machine = { config, ... }: {
imports = [
./common/auto.nix
./common/user-account.nix
];
test-support.displayManager.auto = {
enable = true;
inherit user;
};
services.xserver = {
enable = true;
desktopManager.mate.enable = true;
};
services.displayManager.defaultSession = lib.mkForce "mate";
services.ayatana-indicators = {
enable = true;
packages = with pkgs; [
ayatana-indicator-datetime
ayatana-indicator-messages
ayatana-indicator-session
] ++ (with pkgs.lomiri; [
lomiri-indicator-network
telephony-service
]);
};
# Setup needed by some indicators
services.accounts-daemon.enable = true; # messages
# Lomiri-ish setup for Lomiri indicators
# TODO move into a Lomiri module, once the package set is far enough for the DE to start
networking.networkmanager.enable = true; # lomiri-network-indicator
# TODO potentially urfkill for lomiri-network-indicator?
services.dbus.packages = with pkgs.lomiri; [
libusermetrics
];
environment.systemPackages = with pkgs.lomiri; [
lomiri-schemas
];
services.telepathy.enable = true;
users.users.usermetrics = {
group = "usermetrics";
home = "/var/lib/usermetrics";
createHome = true;
isSystemUser = true;
};
users.groups.usermetrics = { };
};
# TODO session indicator starts up in a semi-broken state, but works fine after a restart. maybe being started before graphical session is truly up & ready?
testScript = { nodes, ... }: let
runCommandOverServiceList = list: command:
lib.strings.concatMapStringsSep "\n" command list;
runCommandOverAyatanaIndicators = runCommandOverServiceList
(builtins.filter
(service: !(lib.strings.hasPrefix "lomiri" service || lib.strings.hasPrefix "telephony-service" service))
nodes.machine.systemd.user.targets."ayatana-indicators".wants);
runCommandOverAllIndicators = runCommandOverServiceList
nodes.machine.systemd.user.targets."ayatana-indicators".wants;
in ''
start_all()
machine.wait_for_x()
# Desktop environment should reach graphical-session.target
machine.wait_for_unit("graphical-session.target", "${user}")
# MATE relies on XDG autostart to bring up the indicators.
# Not sure *when* XDG autostart fires them up, and awaiting pgrep success seems to misbehave?
machine.sleep(10)
# Now check if all indicators were brought up successfully, and kill them for later
'' + (runCommandOverAyatanaIndicators (service: let serviceExec = builtins.replaceStrings [ "." ] [ "-" ] service; in ''
machine.succeed("pgrep -u ${user} -f ${serviceExec}")
machine.succeed("pkill -f ${serviceExec}")
'')) + ''
# Ayatana target is the preferred way of starting up indicators on SystemD session, the graphical session is responsible for starting this if it supports them.
# Mate currently doesn't do this, so start it manually for checking (https://github.com/mate-desktop/mate-indicator-applet/issues/63)
machine.systemctl("start ayatana-indicators.target", "${user}")
machine.wait_for_unit("ayatana-indicators.target", "${user}")
# Let all indicator services do their startups, potential post-launch crash & restart cycles so we can properly check for failures
# Not sure if there's a better way of awaiting this without false-positive potential
machine.sleep(10)
# Now check if all indicator services were brought up successfully
'' + runCommandOverAllIndicators (service: ''
machine.wait_for_unit("${service}", "${user}")
'');
})