Sometimes it's useful to be able to specify multiple references and not
just a single one. When list is provided with this patch per-item
discovery is performed.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Allowing name_prefix to be list allows one to specified shared parts of
the name_prefix inherited from upper-node:

    !mux
    functional: !mux
        name_prefix: ["functional-"]
        rpm:
            name_prefix: ["rpm"]
        git:
            name_prefix: ["git"]

which results in: "functional-rpm" and "functional-git".
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

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

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

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

To fix the pip 9.0.1 install python requirements with ">=2.7, >=3.4" failure,
">=2.7, >=3.4" means ">=2.7" and ">=3.4", so the result is ">=3.4"
for reference: https://stackoverflow.com/questions/44660448/Signed-off-by: NJunxiang Li <junli@redhat.com>

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

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

We are, finally, advertising Python 3 support.  Yes, it's supposed to
be "beta" quality level, as there are a few leftover issues we know
about, and possibly a whole lot more we don't.

Still, we're taking that leap of faith, and that suggests describing
that so that PyPI users and installation tools know about it.

Reference: https://packaging.python.org/tutorials/distributing-packages/#classifiers
Reference: https://packaging.python.org/tutorials/distributing-packages/#python-requiresSigned-off-by: NCleber Rosa <crosa@redhat.com>

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

The API used to get the file, by name, is `get_data()`, which looks
for that file in one of the data directories that the class provides
(file-level, class-level, variant-level, etc).
Signed-off-by: NCleber Rosa <crosa@redhat.com>

These parameters are not class instantiation parameters, which is
the indentend usage of those docstring notations.  But, even worse,
they don't exist in these example tests at all.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

We don't do Python 2.6 anymore, so it's safe to drop this workaround.
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>

The "test" fields was removed from JSON output as it was
legacy/duplicated. This patch adapts the remote runner accordingly,
using the key "id" instead.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

As part of the v59 release process.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Now there's a new parameter executable to use as the commandline
information. The previous parameter, name can now be hijacked by
the yaml_loader, making this plugin to break.

This patch adapts to that change of internal API.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Now there's a new parameter `executable` to use as the command line
information. The previous parameter, `name` can now be hijacked by
the yaml_loader, making this plugin to break.

This patch adapts to that change of internal API.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Now there's a new parameter `executable` to use as the command line
information. The previous parameter, `name` can now be hijacked by
the yaml_loader, making this plugin to break.

This patch adapts to that change of internal API.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

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

This check was introduced in the previous draft of this plugin, when it
was using gtester, and left over. Dropping.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

As reported in the following link:
https://github.com/avocado-framework/avocado/issues/2476
build.run_make fails to run when used without any argument. This
patch handles the failure and runs make with ncpus if it is not
specified in "env"
Signed-off-by: NHarish <harish@linux.vnet.ibm.com>

Signed-off-by: NCleber Rosa <crosa@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>

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

The current code style creates and appends an object into the list,
and then later implicitly references that item using the `[-1]`
notation, changing its filter.

While correct, IMO, it's easier to follow code that creates the
object, sets its values, and then appends it to the list.

This was a style only change that appeared while debugging another
mux related issue.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

When things go wrong, the Avocado command line app will write a file
with the backtrace that caused the trash.  Depending on its content,
it may not be possible to encode that in an file encoded as ASCII.
Let's default to utf-8 then.

Also, this is done for portability reasons, as it brings a consistent
behavior (one explicit encoding type) across Python 2 and 3, which
have different default encodings.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

On Python 3, trying to copy a FilterSet, using the instance.copy()
method, will return a plain set() object.  This influences a number of
code, including the fingerprint generation, and the variant ID which
will be based on the hash of the fingerprint.

Let's make things predictable and copy using Python's standard module
copy.

This fixes the functional test in module `test_getdata` under Python
3, as the variant IDs are now the same on jobs run under Python 2 and
3.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Since system_output() returns bytes (from CmdResult.stdout), the
split operation should also be done with bytes.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

In situations where a user of the CmdResult class manually sets the
stdout/stderr attributes to a string, instead of the recommended and
documented content of type "bytes", it won't be possible to "decode"
the bytes into a string of a given encoding, and we'll end up with
a crash.

Since the goal of std{out,err}_text is to return a text version
of stdout, if itself already is of such a type, let's just return
it.

Additionally, if the data cannot be "decode()d", let's raise an
explicit TypeError on this location, rather than later in the code
when its value is attempted to be used.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Commit 7d0e6a44 effectively changed the supposed value of stdout and
stderr attributes to being of type "bytes".  But, the parameters with
the same names, and their defaults values was not updated.

Let's make stdout and stderr into bytes by default, both the
parameters and attribute values when no other is given.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

On Python 2, when running the standard unittest runner to run Avocado
tests, the garbage collection on the temporary dirs have been
performed, but depending on other code to clean up directories that
are created at class instantiation is not a very good idea.

On Python 3, the standard unittest runner won't clean up the
directories created by the Avocado test instances.

Being initiated on-demand, instead of on class instantiation also
means faster code (no disk writes) at every test class instantiation.

This fixes 3 leakages for temporary dirs when running functional
tests on Python 3 (on module test_unittest_compat).
Signed-off-by: NCleber Rosa <crosa@redhat.com>

On Python 3, the garbage collection doesn't clean up standalone jobs
temporary dir without an explicit __del__ implementation.

This fixes the leakage on 7 directories when running functional
tests on Python 3 (on module test_standalone).
Signed-off-by: NCleber Rosa <crosa@redhat.com>

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

* https://github.com/avocado-framework/avocado:
  Support variants dump/load
  core varianter: implement __len__

Now users can run in Avocado the tests created with the GLib Test
Framework with the same level of granularity as they have when executing
the tests directly.
Signed-off-by: NAmador Pahim <apahim@redhat.com>

Depending on the number of parameters, generating the Variants can be
compute intensive. This patch adds the ability to dump/load the variants
to/from a JSON file so the Variants generation can be offloaded from the
machine under test.

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