TODO


1
2
3
4
5
6
7
8
9
10
11
12

- actual realtime scheduling (maybe). sched(7).
- use a log file instead of stderr, so we can have error/warning messages
  after detaching from the tty.
- find out what happens if we try to run a 2nd instance.
- print our /dev/input/event* node name, if possible.
- options to set the virtual device name and vendor/product IDs.
- much as I hate systemd, get someone to write a systemd unit.
- s6 and/or openrc init scripts?
- possibly use a UNIX socket. if we can connect to it on startup,
  that means another instance is already running... so either exit
  or pass a command to it (like "marsond -K" would kill an existing
  instance).