GAppInfoMonitor documentation deficiencies
Submitted by Matthias Clasen
GAppInfoMonitor is named very similar to GFileMonitor, but it works very differently. With GFileMonitor, you get the object once, and then you can listen for changed signals until you drop the object.
In contrast, changed signals on GAppInfoMonitor are one-shot, and you have to call some g_app_info api to re-arm the monitor.
The documentation does not really spell that out very clearly, although it does hint at the reason for this behaviour: rescanning all applications is expensive. I would expect to see the word 'one-shot', and a list of functions that rearm the monitor.