selftests/functional/test_replay.py: fix typo that caused sysinfo col…

I noticed during a run of the self test suite that system information
was being collected (`rpm -qa` and the like can usually be seen using
quite a few resources).

It looks like this little typo was causing that.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

This reverts commit 732a400a.

avocado.main(): avoid an infinite fork loop bomb [v4]

Because the current implementation of avocado.main() creates a job
and runs "sys.argv[0]" to implement standalone mode, it ends up
running itself over and over.

This simple proposed fix, prevents avocado.main() from running
itself again if called from itself. Since they are on different
processes, the mechanism chosen to do this is to set an environment
variable, that will be seen by the next process.

Also, by exiting from main() with an error code, the test first
level test will fail. This will let the user know that the chosen
approach (SIMPLE tests written in Python and calling main()) are
not worthy of a PASS.

The functional tests make use of Python's standard library utilities
(subprocess module) directly for running Avocado because of current
issues with Avocado's own process utility module.

This adresses issue #961.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Test namespace cleanup [2nd round] [v2]

IMHO it makes more sense to only publish these variables when they
actually contain meaningful value. Right now it *may* contain "None"
as this is maybe the value of the `basedir` and `datadir` attributes
when the test is not backed by a file.

Also, it's a lot more common, say, for a shell script to check if
a variable is set (if [ -n $AVOCADO_TEST_DATA_DIR ] ...) then to
check if it's not equal to "None".
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Avocado doesn't intent to support the use of these attributes by
test (writers), so let's at the very least hide them.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

The implementation of basedir from Avocado Test is also sufficient for
SimpleTest now that the `filename` property is implemented.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

SIMPLE tests are backed by files, executable ones. So, it makes sense
to have the `filename` property accordingly. The base Avocado Test
class implementation of `basedir` applies to SimpleTest as well
with this implementation.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Since basedir is closely related to `filename` and `datadir`, let's
also make it a property for coherency.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

It now becomes a first class citizen, with proper documentation.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Some Test types are not backed directly by files, and thus do
not have a datadir. One example is Avocado-VT's VirtTest class.
Because of that, and because other custom Test types may adhere
to the same logic of not having a datadir, let's make Avocado's
base test class more robust against a `None` datadir.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Now that 7c967bb0 hid the methods that record/check stdout/stderr,
let's also hide the attributes that hold their paths.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Let's just mirror the parameter order of the super class. It's just
easier on the eyes.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Using an unknown option in a given subcommand leads argparse to
show the main avocado usage instructions.

This patch makes the argparse show the subcommand help instead.

Also, argparse error exit code now matches error_codes.AVOCADO_FAIL,
which makes more sense than error_codes.AVOCADO_JOB_FAIL.

Reference:https://trello.com/c/o0TpH9aeSigned-off-by: NAmador Pahim <apahim@redhat.com>

Test namespace cleanup

avocado.core.test: PEP8 Fix

test_interrupt.py: Finish change a1afd91b

Turns out I thought out a1afd91b poorly, I needed to use
a more encompassing try/except block, and replicate the
block in the two tests that do psutil shenanigans.
Signed-off-by: NLucas Meneghel Rodrigues <lookkas@gmail.com>

Signed-off-by: NLucas Meneghel Rodrigues <lookkas@gmail.com>

The reason being that `datadir` can then be properly documented as
first class citizens so that Test class users (test writers) can rely
on it.

Other reason is that that it can be overloaded by inherited classes
such as `SimpleTest`, without the need to (re-)set other attributes
that depend on it.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

The methods `record_reference_stdout`, `record_reference_stderr`
`check_reference_stdout` and `check_reference_stderr` are publicly
visible, but are not intended to be used by test writers and are
not intended to have API stability.

Let's rename them so that they're hidden from the Avocado test
writer.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Currently replay is using status.mapping as the list of valid status
for the --replay-test-status option. But status.mapping has some internal
status that users should not have to deal with.

This patch creates a user facing status which is a subset with only
the options relevant for the users and makes the job replay feature to take
advantage of it.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

This patch is merging the concepts of TEST_NA and SKIP, making
the consequent status more relevant for the users.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Turns out that setting the argument value directly does not suit most
of our code. Let's revert back to the old way of setting the remote/vm
result/runner until we implement proper runner and result plugin interfaces.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

For parity of features with the remote plugin, let's add the SSH
port as an option of the VM plugin as well.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

the replay_parser is only needed during __init__, it does not need to
be stored in self.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

When using replay, one can think of the replay_multiplex as of the
normal multiplex files, only it's already processed. This patch
stores the retrieved multiplex file (if not overridden) directly in
multiplex_files and adds support in avocado.core.job to handle this
situation.

Actually some future plugins might benefit from this and create their
own inherited Mux class instead of our multiplexer.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The replay_chdir is not useful for avocado.core, therefor it should
live within the replay plugin. Unless the user uses some incompatible
plugin which also changes the dir it should work.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

The "make link" target purpose is to help avocado developers to prepare
environment to run avocado from sources including all subprojects.
Nowadays it requires to run "setup.py develop" manually, but let's
assume wast majority of users of "make link" feature use "setup.py
develop --user" and call it automatically for each subproject (if they
support that).

This way simple "make link" prepares the environment to run avocado from
sources. It can also be used when new targets (plugins) are added
without the need to think which subproject needs to be updated.

NOTE: The subprojects are responsible to add the "python setup.py
develop --user" into the "link" and "unlink" targets.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>