The current way of handling errors isn't exactly great:
When an error occurs, we require the end-user experiencing the error to tell the instance administrator when exactly they've seen the error in order to get any information about it. Probably we want to properly setup and use the tracing crate instead.
Only the database part is currently tested in tests, but not the real and whole application. At least a roundabout-test with a locally spun-up server like Rocket offers would be nice, to catch some accidental changes to the registration mechanisms (which is pretty much finished and not to be seriously changed atm).
The current way of handling errors isn't exactly great:
tracingcrate instead.