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

> Blocked on Alan's "Needs Fixing". Not sure what needs fixing though. Looks
> like a general disapproval of the whole concept without any alternate
> solutions for bug 1316867 being proposed yet.

What "Needs Fixing" is:

1. Aborting as the only action for fatal errors.
2. Aborting as the default action for fatal errors.

Both of would be solved by introducing a mechanism that allows the _users_ of libmirserver to decide what to do in case of fatal errors. The default implementation would be to throw an exception, which, as discussed above, is the most safe and polite reaction.

I expect unity8 and USC to override the fatal error handler and abort() on error to provide sufficient debugging information in the wild, but that's ultimately their decision to make.