Signed-off-by: NAmador Pahim <apahim@redhat.com>

The FDDrainer reads from process output and write to loggers.  The
test log, which is intended for humans, can indeed try to treat it as
text.  The stream logger, on the other hand, should deal with data as
binary content *only*, but this will be handled on a future change.

By having text content passed to the loggers, we avoid the occurrence
of log lines that look like:

  [stdout] b'command output'

Or output files (job-YYYY-MM-DDTHH.MM-id/tests/<test>/output) that
may literally contain:

  b'command output'

The encoding used is the one specified at the command execution.  With
this fix, a number of tests on Python 3 can now run successfully.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

In the module global scope, "man echo" is executed to attempt to
determine if it's a GNU echo.  But, "man echo" in a locale such
as UTF-8 produces funky copyright characters:

  $ LANG=en_US.UTF-8 man echo | chardetect -
  <stdin>: windows-1252 with confidence 0.73

Which can not be decoded by the UTF-8 codec, producing
UnicodeDecodeErrors.  Even if that wasn't the case, it's better to
control the locale of command that can generate varied output, and
that we depend on.

  $ LANG=C man echo | chardetect -
  <stdin>: ascii with confidence 1.0
Signed-off-by: NCleber Rosa <crosa@redhat.com>

This change introduces the following requirement: the command to
be run by the avocado.utils.process functions should be given as unicode.
It brings a clearer interface, in which there's no need to deal with
encodings internally.

For simple test, it brings changes to respect the of passing unicode.
And, it changes the presentation of simple tests with unicode names
explicitly expreseed as such.

The way to deal with the process data that may be generated on
stdout/stderr is unchanged.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

There's just so much implicit conversion on those methods that makes
them an enourmous source of errors (including UnicodeErrors).  The
information is still all there, and users are free to interpret them
as they see fit.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

On the basis of the non-existence of the "true" command on the system
running tests.  Most of the tests do not even reference the "true"
command, and those that do, do not require the binary to exist, but
only a valid command line to be given.

In that aspect, a fictional command is used where needed.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

The sudo related test are currently broken in the following aspects:

 1) it doesn't require TRUE_CMD
 2) the monkey patching of find_command() produces and adverse effect,
    and makes the binary we look for ("sudo") become "true"
 3) the expected results reflect the wrong sudo binary

Let's fix those aspects, making sure the process.py library looks for
sudo only, and that it prepends that (the found sudo binary) to the
generated command.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

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

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

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

To lay some common ground across Python versions, let's define
the criteria we have for binary or text data.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

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

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

Since we are not using __dir__ anymore to list the available properties,
the OrderedDict is not needed. And dropping it makes the performance to
improve.

Before this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 1.1071600914001465

After this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 0.9833340644836426
Signed-off-by: NAmador Pahim <apahim@redhat.com>

For small classes, like this one, with known attributes, using dict
wastes a lot of RAM. Defining __slots__ as a list will save a lot of
memory, while improving the performance a bit.

Before this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 1.1383171081542969

After this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 1.1071600914001465
Signed-off-by: NAmador Pahim <apahim@redhat.com>

We loose in flexibility by the performance improvement justifies the
change. Since __getattr__ is the last resource python tries to solve
attributes, creating the actual properties makes the resolution to
happen earlier and the performance to be way better.

Before this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 2.0695440769195557

After this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 1.1383171081542969
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Instead of using regex to parse the `data`, let's use a simpler and more
performatic version witch just handles the last character to figure out
the unit.

Before this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 2.988218069076538

After this patch:

>>> timeit.timeit('data_structures.DataSize("1M").b',
                  setup='from avocado.utils import data_structures',
                  number=1000000)
>>> 2.0695440769195557
Signed-off-by: NAmador Pahim <apahim@redhat.com>

- try harder to detect arm.
- detect more archs (s390, sparc64 and i386).
- fallback to plataform.machine() (instead of returning i386).
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Signed-off-by: NAmador Pahim <apahim@redhat.com>

Signed-off-by: NAmador Pahim <apahim@redhat.com>

Signed-off-by: NAmador Pahim <apahim@redhat.com>

There are few issues (3 to my count) which affect a number of tests
on Python 3.  Instead of not executing any of the functional tests,
let's skip the broken ones.

This is important because it will prevent new issues from making
its way into the Avocado code.

The list of tests being skipped is being tracked on our planning board,
so that they are accounted for, and enabled when fixed.

Reference: https://trello.com/c/eFY9Vw1R/1282-python-3-functional-tests-checklistSigned-off-by: NCleber Rosa <crosa@redhat.com>

This prevents the plugin from trying to run references such as
Avocado-VT tests or any other that is not backed by an executable
file.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NAmador Pahim <apahim@redhat.com>

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

And that have been causing failures on Travis jobs, due to limited
on unreliable amount of resources present.

This is the full list of tests that won't be run anymore on Travis,
only on `make check-full`:

  selftests.functional.test_basic.RunnerOperationTest.test_early_latest_result
  selftests.functional.test_basic.RunnerSimpleTest.test_kill_stopped_sleep
  selftests.functional.test_interrupt.InterruptTest.test_well_behaved_sigint
  selftests.functional.test_interrupt.InterruptTest.test_well_behaved_sigterm
  selftests.functional.test_loader.LoaderTestFunctional.test_simple_using_main
  selftests.functional.test_loader.LoaderTestFunctional.test_sleep_a_lot
Signed-off-by: NCleber Rosa <crosa@redhat.com>

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

Some tests might produce too much logs which might result in too big
xunit results, that might be hard/impossible to publish. Let's allow
setting a limit to number of characters to those files and embed only
1/2 of the limit from the beginning and 1/2 of the limit from the end of
the log.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NAmador Pahim <apahim@redhat.com>

Added support to pass full string of block id like 'memory234',
it saves few steps of the user to find the memory blocks and call
this functions to hotplug/unplug
Signed-off-by: NAbdul Haleem <abdhalee@linux.vnet.ibm.com>

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

The fingerprint is used when replaying jobs causing "yaml_loader" test
not being replayable.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

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

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

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

Waiting 120s is not appropriate as the system takes less time for
memory state to change from 'going-online' to 'online' so changed
it to 10s and with freq of 0.2s
Signed-off-by: NAbdul Haleem <abdhalee@linux.vnet.ibm.com>

When this test was initially written, the concept of a shared
directory available across tests did not exist.  Let's optimize the
execution by avoid compilation of the same binary on every single
test.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

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

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

The newly W605 warning is caused by incorrect escaped chars, that are
automatically translated to '\\' but future versions should treat it as
SyntaxError.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>