Fix-committed: #10160

Fix-committed: #10147

Conflicts:
	run_test_suite

Fix-committed: #9491

It was autoremoved from Debian Testing due to the Debian bugs:

We'll need that to hook the pseudo fifo logging system into the
chatbot helper.

Fix-committed: #9375

kytv authored Jun 22, 2015 and anonym committed Sep 11, 2015

Our otr-bot.py does not use logging but the jabberbot library makes
logging calls,  causing a one-off message “No handlers could be found
for logger "jabberbot"” to be printed to the console. This commit
effectively prevents logging/outputting anything to the terminal which
is at a level lower than CRITICAL.

Will-fix: 9375

We don't want stray debug things to linger from previous runs. We
don't use FileUtils.rm_f since we want to error if there's permission
issues since they otherwise will cause (possibly harder to detect)
errors later.

That step really requires that Tails is running to make sense, so any
scenario using it otherwise is incorrect.

We don't need the interactive progress report written to STDERR, and
we don't want to pollute it any more since we have this cucumber
formatter based logging.

It's always the same (=> useless) any way.

Actually this overwrites cucumber's 'debug' formatter, but it's
useless for anything except developing new formatters, which we might
as well do in a cucumber environment outside of Tails.

Will-fix: #9491

... because CRuby's threading support sucks when IO is involved.

The idea is to replace each instance where we print debugging info to
STDERR if $config["DEBUG"] is set, with a call to debug_log(), which
will make the pretty_debug formatter print the output *if* it is
actiaved.

This approach is cleaner than what we do now, and will make it
possible to e.g. usw the pretty formatter for the terminal output and
at the same time logging pretty_debug to a file. Upon error, the full
log can then be consulted, without making the terminal output
messy. While debugging/developing the test suite, the pretty_debug
formatter can still be directed to the terminal, of course.

A notable problem is how to deal with the various subprocesses/threads
whose STDERR output we also are interested in when debugging only,
like Sikuli's debugging info. We'll have to find a way to capture such
output and hook it into debug_log() some how.

... by moving the extra hooks into a null formatter that we always
activate. This also fixes a bug: if the first formatter was specified
twice (e.g. --format pretty --format pretty --out /tmp/test.log) then
the extra hooks would be run twice.

It's been removed with #9597.

If panic mode failed to disable the specific device that couldn't be
spoofed (by unloading the module) we disable networking. Previously we
only stopped NetworkManager. The problem is that NM isn't even started
at this time, but will specifically be started when we're done with
MAC spoofing.

Therefore, let's remove NetworkManager so it cannot possibly be
started.

Will-fix: #10160

Fix-committed: #9890

Fix-committed: #9877

Using the same ISO for the USB upgrade tests most often still does
what we want, e.g. test that the current version of Tails being tested
has a working Tails installer. Hence this seems like a reasonable
default. (Also note that an --old-iso may have changes that make it
impossible to run in the next versions automated test suite, and then
this is almost a requirement.)

Will-fix: #10147