Code review comment for lp:~vanvugt/mir/log-level

Needs information/discussion.

I think I understand the logic for servers and clients.

# Servers: Log everything
# Clients: Log errors only by default, or more if a report requires.

A server is unlikely to be run directly by the user, or if it is, is likely to be run by a user
with an interest in log output. Likewise, clients are relatively likely to be run by a user, e.g.
running an app from a terminal.

I don't think I agree with the tests case though:
# Tests: Log errors only

It's easy to see if a test fails and locate the failing test. Either by color coding or searching for "FAIL". All passing test output can be ignored. Why else wouldn't you want the most detailed logging possible for a failing test?