Code review comment for lp:~vanvugt/mir/fatal-error

I wonder if there is a way we can recover the VT/display using a separate executable tied to a watchdog, or a hot-key combination of sorts which can be triggered automatically or through a hot-key combination, to get the display back into a sane state after a fatal crash.

And perhaps a good history logging mechanism, as suggested by Chris, which we can "play back" post-mortem in a developer binary where we have aborts instead of exceptions.

I am also of the opinion that the user-experience must not be compromised just because we have bugs in the system we want to address as quickly as possible. And there will continue to be bugs even after we collect many core files and address the ones that have caused them. But we should not forget that providing a good user-experience is key - piss your users off enough many times and there won't be very many left to exercise bugs in the code and help us collect core files for us to fix.