Comment 1 for bug 1074993

This is still a fairly serious problem for libfolks, since it’s causing spurious failures in our test suite, and I can find no way to work around it.

On further digging, there’s a second (separate) problem which is triggered at the same time: the monitor isn’t removed from the ZeitgeistLog’s list of monitors until the async D-Bus call returns. Calls to zeitgeist_log_remove_monitor() are predicated on whether the monitor is present in the list of monitors. Therefore it’s possible for multiple D-Bus calls to be started before the first one returns.