pytest removed copy() method from its Namespace class. Use the copy
module to make a copy of early options.
Signed-off-by: Christian Heimes <cheimes@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
The Python 3 refactoring effort is finishing, it should be safe
to turn all scripts to run in Python 3 by default.
https://pagure.io/freeipa/issue/4985
Reviewed-By: Christian Heimes <cheimes@redhat.com>
ipa-run-tests now applies the same logic to --ignore then to included
paths.
https://pagure.io/freeipa/issue/7355
Signed-off-by: Christian Heimes <cheimes@redhat.com>
Reviewed-By: Michal Reznik <mreznik@redhat.com>
The ipa-run-tests command used os.chdir() to change into the ipatests/
directory. The approach works for simple cases but breaks some pytest
features. For example it makes it impossible to selects tests by their
fully qualified test name.
Further more, coverage statistics break because path and module names
get messed up by chdir.
A name plugin takes care of adjusting paths relative to ipatests and to
add ipatests as base. It's now possible to run tests with qualified test
names, e.g.
ipa-run-tests ipatests/test_ipalib/test_base.py::test_ReadOnly::test_lock
Signed-off-by: Christian Heimes <cheimes@redhat.com>
Reviewed-By: Felipe Volpone <fbarreto@redhat.com>
Closes: https://fedorahosted.org/freeipa/ticket/6631
Signed-off-by: Christian Heimes <cheimes@redhat.com>
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
Let's replace some ugly hacks with proper pytest conftest.py hooks.
Test initialization of ipalib.api is now handled in
pytest_cmdline_main(). Pytest plugins, markers and ignores are also
moved into conftest.py. Additional guards make it possible to run tests
without ipaserver installed.
I added confcutdir to ensure that pytest does not leave our project
space. Pytest used pytest.ini or setup.py before but pytest.ini is gone.
Signed-off-by: Christian Heimes <cheimes@redhat.com>
Reviewed-By: Milan Kubik <mkubik@redhat.com>
Logs from Beaker jobs are normally very brief, with the standard
output/error containing detailed information. Make ipa-run-tests
with BeakerLib plugin follow this convention.
Only include INFO and higher level messages in the Beaker logs.
Downgrade several message levels to DEBUG.
Log to console using Python logging instead of showing the Beaker logs.
Since ipa-run-tests sets up its own logging, Nose's own log
handling just causes duplicate messages. Disable it with --nologcapture.
Output from IPA's log manager is not captured by Nose's logcapture plugin.
Forward IPA logs to a regular Python logger so that they are shown
on failures.
IPA log messages are also shown on standard error.
Filter out Paramiko logs by default; these are too verbose.
Part of the work for: https://fedorahosted.org/freeipa/ticket/3621
Tests in test classes decorated by @ipatests.order_plugin.ordered
are sorted by the source line number instead of alphabetically,
if the plugin is enabled.
The ipa-run-tests helper now loads and enables the plugin.
This should make writing integration tests easier.
The plugin hooks into the Nose runner and IPA's logging infrastructure
and calls the appropriate BeakerLib functions (rl*).
IPA's log_manager is extended to accept custom Handler classes.
The ipa-run-tests helper now loads the plugin.
Patr of the work for: https://fedorahosted.org/freeipa/ticket/3621
