Mailman 3 April 2020 - pytest-dev

background talks on pytest happenings
by holger krekel April 18, 2020

April 18, 2020

Hi pytest folks, i am still talking with various people involved with pytest happenings (culminating in several core devs taking out-time) last week. I got involved after some folks asked me, as founder and 10y+ maintainer of pytest, to help find a way out. If you have anything you'd like to tell me, please write me 1:1, especially if you have comments or suggestions on how to proceed. I aim to have a suggestion later in April. Please bear with me in taking some time as i am also dealing with a lot of non-pytest and personal issues currently, and it took and takes time to schedule talks. Meanwhile, thanks to all the people who did such a great job with pytest the last years and sorry for all the bad code i've written and that couldn't be mitigated yet ;) thanks and best, holger

1 0

Re: [pytest-dev] How to duplicate pytest.Item inside function 'pytest_collection_modifyitems'?
by Ronny Pfannschmidt April 15, 2020

April 15, 2020

*copey with the correct mail* Hi Shiban, its not safely possible to clone a item in general. A) you need to make a new nodeid to prevent some issues B) there is no safe cloning mechanism declared, different subtypes of items have different mutable state you dont want to mix I believe the closest you can get is creating a new item that is reasonably equivalent. The Details of that would depend on what the actual Item subtype is. -- Ronny Am 15.04.20 um 07:45 schrieb Ronny Pfannschmidt: > > Hi Shiban, > > its not safely possible to clone a item in general. > > A) you need to make a new nodeid to prevent some issues > > B) there is no safe cloning mechanism declared, different subtypes of > items have different mutable state you dont want to mix > > I believe the closest you can get is creating a new item that is > reasonably equivalent. > The Details of that would depend on what the actual Item subtype is. > > -- Ronny > > Am 14.04.20 um 23:46 schrieb Anton Shiban: >> >> Hi, >> >> >> Can you please guide me *how to* *clone pytest.Item?* >> >> I want to take one *item* from *items* array (inside function >> ‘pytest_collection_modifyitems’) and clone that *item* and add it to >> *items* array back. >> >> Can someone please guide to do this? >> >> >> Thanks, >> >> Shiban. >> >> >> _______________________________________________ >> pytest-dev mailing list >> pytest-dev(a)python.org >> https://mail.python.org/mailman/listinfo/pytest-dev

2 1

How to duplicate pytest.Item inside function 'pytest_collection_modifyitems'?
by Anton Shiban April 15, 2020

April 15, 2020

Hi, Can you please guide me *how to* *clone pytest.Item?* I want to take one *item* from *items* array (inside function ‘pytest_collection_modifyitems’) and clone that *item* and add it to *items* array back. Can someone please guide to do this? Thanks, Shiban.

3 3

also leaving pytest-dev for now
by Ronny Pfannschmidt April 12, 2020

April 12, 2020

Hi everyone, after contemplaing a while i decided to join Bruno and Anthony. When i returned from my hiatus i declared that joy will be important, currently that joy is absent. -- Ronny

5 6

Code of conduct issues
by Brianna Laugher April 11, 2020

April 11, 2020

Hi all, As one of the Code of Conduct (CoC) committee members ( https://github.com/pytest-dev/pytest/blob/master/CODE_OF_CONDUCT.md ) I want to apologise for the poor handling of issues that have led Bruno, Anthony and Ronny to step back from the project. As a group we failed to act decisively and now people who have contributed so much to pytest are suffering, and for that I am sorry. I intend to reach out to the PSF Conduct Working Group and ask them to advise us on a way forward. thanks, Brianna

1 0

multiprocessing tests hang on Python3.8+macOS
by Andrew Nelson April 11, 2020

April 11, 2020

Dear pytest-dev, on scipy we typically run the test suite using a test script collecting a series of test arguments, then supplies them to `pytest.main()`. https://github.com/scipy/scipy/blob/master/runtests.py#L307 then https://github.com/scipy/scipy/blob/master/scipy/_lib/_testutils.py#L67 I'm finding that all the tests that involve multiprocessing fail, e.g. ``` def test_pool(): with Pool(2) as p: p.map(math.sin, [1,2,3, 4]) ``` This is on macOS+Python3.8 that uses 'spawn' as a start method. If I run the tests directly using pytest there aren't any issues. Are there any known issues using `pytest.main` with multiprocessing? Andrew

1 1

Also leaving pytest-dev for now
by Anthony Sottile April 10, 2020

April 10, 2020

Hello friends, Also echoing the concerns of Bruno I'll be stepping down from my role as core developer on pytest. I love working on the project but cannot continue to do so when working with a particular individual has been extremely draining. I, in good faith, cannot continue to represent a project which enables this individual's behaviour towards other contributors and users. So until further notice, I'll be taking leave from the project. Anthony

1 0

Leaving pytest for now
by Bruno Oliveira April 10, 2020

April 10, 2020

Hi everyone, For the past years I've been part of pytest, which is an awesome library and also a community with awesome people. It definitely made me a better person and better programmer. However, the past months or so one individual has drained my will to work on the project because of his interactions, so I decided it is best that I leave the project for a while and focus on other things. I of course wish the project and everyone involved the best, and will definitely continue to work on my numerous plugins meanwhile. Stay safe, and cheers! Bruno

1 0

What's the best way of writing tests for terminal width resize behaviour in an assertrepr plugin?
by Harry Percival April 9, 2020

April 9, 2020

cf discussion here https://github.com/pytest-dev/pytest/issues/162 I'm working on pytest-icdiff, trying to get it to print out well optimised messages for different terminal sizes, and it's hard. when i'm running tests using runpytest, and I _think_ in real life too, config.get_terminal_writer().fullwidth *inside the hook* is returning incorrect values: resizing the terminal i'm using to run the tests has no effect on the value i get from it. and when I use that value irl, real test output does not adjust correctly to the terminal size but calling py.io.TerminalWriter().fullwidth _outside_ the hook, eg at module level, _does_ give me readings that look right. but testing that is hard. intriguingly, if I do testdir.monkeypatch.setenv('COLUMNS', '50'), then the fullwdith *inside* the hook varies. but that's no use to me, since IRL that value doesn't work. so my question is: a) does anyone know of any good examples of assertrepr plugins that adjust dynamically to terminal width, and b) what's a good way of writing tests for that? here's the repo fwiw https://github.com/hjwp/pytest-icdiff/blob/master/tests/test_pytest_icdiff.…

3 3

junitxml incremental output writing
by Floris Bruynooghe April 8, 2020

April 8, 2020

Hello all, One of the most common things users of pytest-timeout struggle with is that they lose Junit XML reporting when a timeout occurs. This never bothered me, it was by design, but that doesn't stop users from being bothered. Which made me wonder, does anyone know if there's a reason that the Junit XML reporting can't write it's report incrementally as each test executed? Not sure that would work well with xdist though. An additional thing or alternative would be if there's a hook that we could use from plugins to say "finish reporting now". So e.g. pytest-timeout could report a failure for the test which timed out, call this hook and only then terminate the process. I'm only bouncing around ill-thought through ideas around this to see what other people think. Cheers, Floris

2 4