Given the last CI jobs, RunnerSimpleTest:test_kill_stopped_sleep() has
been failing often. The obvious reason, given the nature of the test
and the heavily shared and loaded computing resources they run on, is
the lack of resources.

One example of such as failure:

   ======================================================================
   FAIL: test_kill_stopped_sleep (selftests.functional.test_basic.RunnerSimpleTest)
   ----------------------------------------------------------------------
   Traceback (most recent call last):
     File "/home/travis/build/avocado-framework/avocado/selftests/functional/test_basic.py", line 549, in test_kill_stopped_sleep
       % proc.get_output())
   AssertionError: Avocado process still alive 1s after job-timeout:
   JOB ID     : 5c65a06b4ab15e69d6c4a1fc762368e0e99cbc1f
   JOB LOG    : /tmp/avocado_selftests.functional.test_basicHIyhV4/job-2016-03-09T15.08-5c65a06/job.log
   TESTS      : 1
    (1/1) /bin/sleep 60:
   ctrl+z pressed, stopping test (5046)
   ERROR

So, let's give 4 extra seconds for the process to be
terminated. Hopefully failures in the timeout handling will still be
revealed given the relatively small buffer.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Trying to follow the docs to install avocado from source,
make requirements command fails doe to the missing
dependencies.

This patch adds the two missing dependencies I had to install
in addition to the listed in docs.

Reference: https://trello.com/c/1ryTxsXwSigned-off-by: NAmador Pahim <apahim@redhat.com>

Record the args and reload loader options on job replay.

Reference: https://trello.com/c/N3UCnR05Signed-off-by: NAmador Pahim <apahim@redhat.com>

data_dir: make settings dynamic

The configurations related to the "data_dir" are currently read at
module load time. This means that changes in the configuration between
module load time and changes to the settings (such as parsing an
additional config file) is never seen.

Since the avocado command line application parses extra configuration
files (given with `--config`) after the module is loaded, these extra
configuration files are never applied to the data_dir configuration.

There's one more issue with regards to the the settings usage: because
once references to the `settings` instance singleton are grabbed, they
will always point to the same settings objects. The data_dir unittests
exercise changes to the settings objects by replacing that
instance. So, let's refer to the settings using the full location
(module.attribute), which will always give back the "current" (be it
original or replaced) settings instance.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

The current implementation of _get_bogus_settings() solves a partial need
of the setUp() method. Let's combine both into a single (private) utility
method (_get_temporary_dirs_mapping_and_config()) that can be re-used.

The major goal is to re-use it when testing how data_dir reacts to changes
in the settings at run time.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

We do use default to None everywhere. external-runner-chdir seems to be the only
exception, being defaults to 'off'. This was harmless until we need to record
external-runner-chedir to further retrieve it in replay jobs.

This patch changes external-runner-chdir default from 'off' to None.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Asset fetcher changed the fallback directory to be the data_dir/cache
isntead of a test temporary directory. This patch reflects in docs
that change.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

This check "avocado -v" produces correct output and returns 0.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The `--silent` argument is valid for all commands, let's add it to
the core avocado args. For backward compatibility keep supporting it in
"run" subcommand too.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

This patch assumes "silent" arg has the highest priority and it silents
all output including the one enabled by env variables. The reason
is this handling is cleaner and it allows us to enable the root
logger, which is used by our utils and virttest.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

This adds option --store-logging-stream, which allows users enable
logging to file $JOB_RESULTS/$STREAM.$LEVEL. It will not create those
files in $TEST_RESULTS directory!
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The "View" concept was developed to abstract the messages and events.
The problem is it was only abused to pass the status to other places, to
colorize the messages and to allow creating paginated view.

There already is a nice class hooked to all important events, the
"TestResult".

For (not only human readable) messages this commit uses the standard
python logging as it's pretty well known, widely used and very scalable.
The colored output is handled by already existing
"ProgressStreamHandler", which maps: DEBUG,INFO,WARNING and >=ERROR
messages to previously existing: minor, message, warning and error event
types.

The paginator was unified and is initialized during logging reconfigure.
During reconfigure all previously logged messages are re-logged into the
output so one does not lose those messages. Another great difference is
that the Paginator is cleaned at exit by avocado and does not require
complex handling to avoid broken console. To use paginator one just
enables it in "args" and writes to any available stream/stdout/stderr
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The "args.show" is used later in job. If the value is changed by
other arguments (like --silent) it needs to be updated, otherwise
the job can't react it properly.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

This contains some import cleanups, few docstring updates and escaped
new line fixes.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

By default we only enable >ERROR or ==INFO levels to be used with the
ProgressStreamHandler. This limits the possibilities for future
improvements. Let's switch it to >=WARN (stderr) and <=INFO (stdout).
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Both "stream" and "args" are present in the job. Let's only pass
the "job" object and let the Result class to take the needed objects.

Additionally allow overriding the output for json/xunit/html in order
to allow users to specify different value than the one from args.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The logging stream can be optionally followed by ":$LOG_LEVEL" either
using number, or the symbolic python logging level (INFO).
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

When avocado crashes before the "parser.start()" is finished,
"parser.args" is None and we fail to reconfigure the logging. This patch
initializes empty Namespace in such case.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The TermSupport detects compatible terminal to enable or disable colors.
Let's extract this function to make the initialization easier to read
and also make this function available to the whole framework.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The --log is a bit confusing, let's use --show to emphasize we're
dealing with the console output.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

When avocado fails to update the link to the latest results, the whole
job is aborted. This is unfortunate and after a discussion we considered
this as a "soft" issue.

With this patch we only log an information, that the latest link was not
updated into the job-log and proceed to testing.

This commit also extends the link creation to be
concurrent-execution-safe by creating `latest-$PID` symlink first and
then renames the symlink to `latest`. The rename is atomic operation and
does not suffer the delete->create problem, where other process creates
the file before this one.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Safe Loader: a second, more abstract and reusable implementation [v2]

Find for files in multiple locations, caching it when successfully fetched.
Available as fetch_asset() method in avocado.Test().

Reference: https://trello.com/c/KTeMIx0uSigned-off-by: NAmador Pahim <apahim@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Python 2.6 does not support --user in `python setup.py develop`. Let's
use system-wide develop when link/unlink is executed with python 2.6.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Test namespace cleanup [4th round]

The get_data_path utility API of the Test class is a convenience,
it simply joins the name of a given file to the datadir location.

IMHO the value is low when compared to the pollution of the namespace
and yet another compatibility/stability promise.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

The supported interface for getting the test tagged name is the
attribute `tagged_name`, so there's no need to make this utility
method visible on the API.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

These two attributes are simply utility logger instances that test
writers can use to write "directly" to the test process' STDOUT and
STDERR.

I agree that having those ready to use inside a test is a nice to
have, but I don't think it justifies the polution to the test class
namespace.

Let's remove those and document how the user can use Python's standard
logging API to retrieve the logger instances that Avocado provides to
tests. This will continue to exist and be supported.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

The default tag in the class constructor already is None, so no
there's no need for the conditional assignment here.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

The Avocado Test class inherits from unittest.TestCase, and for that
matter it already contains implementations of setUp and tearDown.

It could be argued that the point is to better document those, but
then the same logic should be applied to all other unittest.TestCase
methods.

So, given that argument, let's remove those.
Signed-off-by: NCleber Rosa <crosa@redhat.com>