Improve testing for Monitor and MonitorObserver

[mulkieran]

No description provided.

[mulkieran]

In particular, is it possible to find out useful things by monitoring real udev events on real systems?

At the very least, make sure it can run w/out raising an exception.

Better, deduce a property from the sequence of udev events observed.
For example, if a device was added and not subsequently removed, it must exist.
It is possible to force such events in software by, e.g., creating and destroying md arrays or what not.
But, this requires imagining a single test, and deciding what the outcome would be, i.e., it is example based.
What about running a continuous test that simply monitors udev events until it finds something that doesn't match expected state? Note that this could not be a hypothesis test, but rather a daemon of some sort.

It is possible to use udevadm monitor as an oracle of sorts.

[mulkieran]

The best first step would be to use udev-hypothesis to simply monitor events as they occur.
The problem, of course, is that they won't occur unless they are forced in some way.

[mulkieran]

We have udevadm trigger and could possibly take advantage of it.