Mailman 3 November 2016 - pytest-dev

t-shirts and stickers now out, thanks Florian!
by holger krekel Nov. 29, 2016

Nov. 29, 2016

Hi all, Florian and Regina visited last weekend, did most of the work on sending out shirts and stickers for both pytest and qutebrowser. Many thanks to Florian who spend almost a week alltogether preparing the shirts, stickers, figuring out and mostly doing all the logistics. I just brought the last packages to the post office and so everything should be on its way now. I'll leave it to him to tell about things to watch out for next time and summarize the experience. After the last costs are claimed i expect there to be around 4-5K EUR left of the original funding FYI. It's not my money so feel free to discuss how to use it. If nothing else it probably can be used for the next sprint or so. best, holger P.S.: Could someone help with setting up appveyor/windows-testing for tox's new https://github.com/tox-dev/tox repository? Would really appreciate it.

7 12

process for replacing yield tests with modern style
by Chris Dent Nov. 28, 2016

Nov. 28, 2016

Several years ago I developed a relative simple YAML-based way of doing HTTP tests. The processed turned yaml into tests and yielded them for pytest to collect. At the time yield was the pytest approved way of doing such things. That's here: https://github.com/tiddlyweb/tiddlyweb/blob/master/test/http_runner.py A few years later I took that idea and turned it into gabbi[1]. Because of local constraints (OpenStack) gabbi needed to be unittest, subunit and testrepository compliant and oriented. Only later, and in rather a hurry, did I add pytest support, using the yield style that I already knew. It's a bit of hacky mess because the tests are ordered unittest.TestCases inside a unittest.TestSuite that has fixtures in the scope of the TestSuite. Each yaml file becomes a TestSuite. For the pytest side of things, plugin code[2] uses the modifyitems hooks to get the fixtures working effectively. Some "driver"[3] code yields tests to a user-code[4] for loading the tests. I recognize that this is all a bit messy, but it does the job. Now, with the advent of pytest 3.x yield tests are deprecated and I'm not sure of the correct way to get the same behavior in modern-style. I'm aware of solutions for using yaml files themselves as the test files[5] but that would be a problem for existing deployed tests that use the style shown in [4] and would also require conftest.py or command line arguments and what I'm hoping to maintain is a way for unadorned discovery to work and for the gabbi tests to be fairly integrated with other tests. I can certainly migrate to the [5] style but I see this as a good opportunity to learn more of the details of how pytest generates tests and use it. Can anyone give me some pointers please? Thanks very much. [1] https://gabbi.readthedocs.io/ [2] https://github.com/cdent/gabbi/blob/master/gabbi/pytester.py [3] https://github.com/cdent/gabbi/blob/2bd1c803b71bd26be7822be87e249a811467b86… [4] https://github.com/cdent/gabbi/blob/master/gabbi/tests/test_gabbits_pytest.… [5] http://docs.pytest.org/en/latest/example/nonpython.html#a-basic-example-for… -- Chris Dent ┬─┬ノ( º _ ºノ) https://anticdent.org/ freenode: cdent tw: @anticdent

4 7

Remove me from this list, my email: eacosta1976@gmail.com
by Edgar Acosta Nov. 24, 2016

Nov. 24, 2016

2 1

pytest-leaks
by Florian Schulze Nov. 24, 2016

Nov. 24, 2016

Hi! I tried pytest-leaks with devpi-server and it seems the way tests are rerun cause lots of failures there. I get errors like "Threads can only be started once" for session fixtures that use threads etc. Not sure if that can be fixed, or if fixing it would interfere with looking for leaks. Regards, Florian Schulze

1 0

Session state
by Alexander Belopolsky Nov. 22, 2016

Nov. 22, 2016

Dear Pytest-Dev, I am trying to implement a pytest plugin to check for reference leaks similarly to how it is done in CPython with python -mtest -R : command. The problem is that pytest_runtest_protocol seems to update some internal state and calling it even on an empty test function increases the total reference count. Clearly, I need to clean up the internal state in my plugin, but I cannot figure out where this state is kept. Looking at the pytest_runtest_protocol implementation in runner.py, I see the following code: def pytest_runtest_protocol(item, nextitem): ... runtestprotocol(item, nextitem=nextitem) return True where runtestprotocol() is a function returning a list to reports, but this list is apparently discarded by the runner. Can someone explain how reports are passed from the runner hooks to the terminal hooks?

2 5

Commit access to pytest
by Floris Bruynooghe Nov. 18, 2016

Nov. 18, 2016

Hi all, A while ago Ronny proposed to adopt the ZeroMQ C4.1 process. While the discussion there never got very far (and I forgot to pick it up at the sprint) I'd like to propose a rather less radical workflow while attempting to make it easier for people to get on board. Basically I'd propose to give commit access to anyone who created a PR and followed it through to it being successfully merged (with changelog etc etc all done by the new contributor). The main reason to propose this is because right now we don't really have a clear policy on when someone joins the committers. And it is much more inclusive to write down clear rules for this. Gaining commit access is generally empowering and motivating and a good way to include new members in the development. What do other people think? Floris

4 9

Fwd: Error in conftest cache
by Dan Nealschneider Nov. 16, 2016

Nov. 16, 2016

I believe that there is a bug in the conftest cache in the Session object. In some situations, conftests are applied based on the last one imported, rather than directory hierarchy. The pytest hook docs say "Session and test running activities will invoke all hooks defined in conftest.py files closer to the root of the filesystem." (pytest docs <http://pytest.org/2.2.4/plugins.html#conftest-py-local-per-directory-plugins> ). I've specifically had problems with creating a new hook, and with the pytest_ignore_collect hook, as I reported here: https://github.com/pytest-dev/pytest/issues/2016. I notice that if I disable the cache (https://github.com/pytest-dev /pytest/blob/master/_pytest/main.py#L568), both of these hooks will operate only on subdirectories, as I would have expected. Another option would be to clear the session's cache each time a new conftest is added (here: https://github.com/pytest-dev/pytest/blob/master/_pytest/config.py#L342), but I am worried about that approach because I know that pytest-xdist has its own session object. Is this the intended behavior? Or does this sound like a real bug? Thanks- Dan W.

1 0

ANN: pytest-bpdb is now part of pytest-dev org on GitHub
by Sławek Ehlert Nov. 14, 2016

Nov. 14, 2016

Just wanted to say hi and mention that pytest-bpdb plugin is now a part of pytest-dev organization on GitHub. https://github.com/pytest-dev/pytest-bpdb Kind regards Sławek Ehlert

1 0

pytest 3.0.4 released
by Bruno Oliveira Nov. 11, 2016

Nov. 11, 2016

Hi all, pytest 3.0.4 has just been released to PyPI. This release fixes some regressions and bugs reported in the last version, being a drop-in replacement. To upgrade:: pip install --upgrade pytest The changelog is available at http://doc.pytest.org/en/latest/changelog.html . Thanks to all who contributed to this release, among them: * Bruno Oliveira * Dan Wandschneider * Florian Bruhin * Georgy Dyuldin * Grigorii Eremeev * Jason R. Coombs * Manuel Jacob * Mathieu Clabaut * Michael Seifert * Nikolaus Rath * Ronny Pfannschmidt * Tom V Happy testing, The pytest Development Team

1 0

Re: [pytest-dev] T-Shirt has arrived!
by Tom Viner Nov. 4, 2016

Nov. 4, 2016

Aha, thanks for explaining! And what a splendid idea Vasily. On 4 November 2016 at 13:10, <pytest-dev-request(a)python.org> wrote: > Yes, indeed the print is supposed to be on the back. It was Brianna's > suggestion to make it so. The thing is, when a girl is wearing a t-shirt > with some elaborate writing on the front, it gets awkward when people try > to read it. I totally sympathise with that. Printing the test results on > the back, on the other hand, allows anyone to discreetly assure themselves > that your garment is working correctly ;) > > Cheers, > Vasily > > P.S. The disadvantage of the print on the back, though, is that I can't > see > your faces in the pictures... perhaps we should have another sprint to take > care of this. >

1 0