Open
Description
Recently we ran this test from ibus 1.5.32 on freebsd, and we noticed that this test is failing.
The log says
# DEBUG: start
# DEBUG: request name returned 1:
# DEBUG: got ownership
# DEBUG: ibus_bus_request_name_async_finish: OK
# DEBUG: name_owner_changed_cb: bus=org.freedesktop.IBus.IBusBusTest, old=, new=:1.3
# DEBUG: ibus_bus_name_has_owner_async_finish: OK
# DEBUG: ibus_bus_name_get_name_owner_async_finish: OK
# DEBUG: ibus_bus_release_name_async_finish: OK
# DEBUG: name_owner_changed_cb: bus=org.freedesktop.IBus.IBusBusTest, old=:1.3, new=
# DEBUG: ibus_bus_add_match_finish: OK
# DEBUG: ibus_bus_remove_match_finish: OK
# DEBUG: ibus_bus_current_input_context_finish: OK
# DEBUG: ibus_bus_list_engines_finish: OK
# DEBUG: ibus_bus_list_active_engines_finish: OK
# DEBUG: ibus_bus_get_use_sys_layout_finish: OK
# DEBUG: ibus_bus_get_use_global_engine_finish: OK
# DEBUG: ibus_bus_is_global_engine_enabled_finish: OK
# DEBUG: ibus_bus_set_global_engine_finish: OK
# DEBUG: ibus_bus_get_global_engine_finish: OK
# DEBUG: ibus_bus_preload_engines_async_finish: OK
# DEBUG: finish_get_address_async: OK
# DEBUG: finish_get_current_input_context_async: OK
# DEBUG: finish_get_engines_async: OK
# DEBUG: finish_get_prop_global_engine_async: OK
# DEBUG: finish_set_preload_engines_async: OK
# DEBUG: ibus_bus_exit_finish: OK socket file: monitored
# IBUS-DEBUG: _connection_closed_cb: Underlying GIOStream returned 0 bytes on an async read
# DEBUG: IBus socket file is deleted
# DEBUG: IBus socket fil
544E
e is created
# DEBUG: IBus socket file's status is 1
#
# DEBUG: name_owner_changed_cb: bus=:1.2, old=, new=:1.2
# DEBUG: name_owner_changed_cb: bus=org.freedesktop.IBus.Config, old=, new=:1.2
not ok /ibus/async-apis - FATAL-ERROR: start_exit_async() is timeout. You might run ibus-daemon with systemd under GNOME and the exit API does not work. You need to export IBUS_DAEMON_WITH_SYSTEMD=1 .
Bail out!
Trace/BPT trap
FAIL ibus-bus (exit status: 133)
My understanding is that this test restarts the ibus daemon, and expects the socket file to change.
What happened, as you can see from the log, was that the file was first deleted, then created. This does not satisfy the test's expectations.
My question is, is this behavior actually acceptable? To me it seems reasonable - the daemon is restarted, so the file is first deleted then created - so should we accept that in the test? or you really expect the file to be modified only?
Metadata
Metadata
Assignees
Labels
No labels