QA » openQA Project » openQA Tests

• Can we import Perl libraries? or do we have to start from scratch and do everything, like ensure_unlocked desktop => Libraries should be exposed for python as well. If not, we should open a ticket. See: https://kalikiana.gitlab.io/post/2021-07-15-writing-openqa-tests-in-python/#fn:1
• How difficult is to write the test (1 to 10) => Can't answer here, since it is subjective and mainly depends on your python knowledge I would say. Might edit this when close to close the story
• Could it be that in the future, as a follow up, we can look into pytest? => Sure, why not? Would be nice to have unit tests for our test code. Should make our tests more stable.
• What could be a proposed structure from an object-oriented perspective? => I expect the test structure to be the direct translation of https://kalikiana.gitlab.io/post/2021-07-15-writing-openqa-tests-in-python/#how-do-i-translate-camel-to-parseltongue Then of course internal methods can use python specific features

• Conclusions after the experiment => TBD

• loadtests mechanism uses python/tests/$squad as default => This does not work at the moment, I would propose to use tests/python/$squad to be compatible without big structure changes

• library should be self contained to python/lib when possible => Same logic as above, lib/python/..

• Optional: A test that uses ssh locally (ssh root@localhost) and runs some commands are written in python => WIP

• Optional: Write a test for Open Broadcaster Software (boot installed system, install obs, open obs, check app starts, close, shutdown the machine) on Tumbleweed, schedule in its testsuite. => ToDo

• Findings are documented in this ticket, with all pull requests related to it => ToDo

• Above questions were answered (during the Sprint review)

Because it was mentioned nowhere in this ticket: Please be aware that the official documentation for openQA lives in http://open.qa
The section
http://open.qa/docs/#_test_case_examples
in particular covers writing openQA test modules in python with example code snippets.

if memory serves well, this is the thing being used: https://metacpan.org/dist/Inline-Python/view/Python.pod

that's correct

Created https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/17325 to continue discussion

szarate wrote:

There's a conversation at: https://suse.slack.com/archives/C02CANHLANP/p1686649742068749

Named argunments are not supported apparently:

perl.require("x11utils")
perl.x11utils.x11_start_program("flatpak run com.obsproject.Studio")

I would consider named parameters being a blocker for being able to use this thing.

 perl.x11utils.x11_start_program("flatpak run com.obsproject.Studio", 30, 1, "flatpak-obsproject") // Ignores other arguments
or
 perl.x11utils.x11_start_program("flatpak run com.obsproject.Studio", target_match="flatpak-obsproject") // Fails to even call the program, tries to call "ARRAY(..."

I think you found a way with perl.x11utils.x11_start_program("flatpak run com.obsproject.Studio", "target_match", "flatpak-obsproject"), right?

Sorry @okurz I missed to answer your question. Yes you are correct, thanks to Sebastian in the discuss-perl channel!

Now PR is merged and it should schedule on the TW Dev job group: https://openqa.opensuse.org/group_overview/38 with job name extra_tests_gnome_development
All was setup yesterday but unfortunately we don't have a TW build today :(

We have it now: https://openqa.opensuse.org/tests/3411838

PR to enhance docs was merged https://github.com/os-autoinst/openQA/pull/5253#event-9874509079

PR documenting further how the python backend works: https://github.com/os-autoinst/openQA/pull/5333

openQA in openQA test provides the simplest python test example: https://github.com/os-autoinst/os-autoinst-distri-openQA/blob/master/tests/openQA/search.py

• There are several limitations already documented in multiple pull requests and in poo#09891
• See http://open.qa/docs/#_notes_on_the_python_api and the how to write tests guide to be able to follow through
• Porting Perl tests to Python would require an immense effort, it can only be used for new test modules that aren't complex (i.e requiring too many things from the perl api.
• Writting tests is possible and loading them is possible too, both: via the schedule and via the main.pm facility, see the openQA in openQA example https://github.com/os-autoinst/os-autoinst-distri-openQA/blob/master/tests/openQA/search.py