[WIP/RFC] fixtures: register finalizers with all previous fixtures in the stack (take 2) #6551

blueyed · 2020-01-23T19:00:22Z

Picking up #6438 again.

TODO:

test for fix using tuple (tests: add test_fixture_arg_ordering #6554)
test for [WIP] Fix exponential recursion #6504 (integrated)
investigate other issues
typing?

/cc @SalmonMode

NOTES:

Sphinx was also affected (Test with pytest-5.3.2 sphinx-doc/sphinx#7037), but worked with this PR here.

Ref: pytest-dev/pytest#6551

This is a regression test for part of pytest-dev#6492, testing one of the fixes in pytest-dev#6551.

blueyed · 2020-01-24T19:24:53Z

@SalmonMode any idea/plan to test #6504 (83349d7)?

SalmonMode · 2020-01-24T19:34:45Z

@blueyed I have an idea. I was working on a test for #6492 / #6551 but I just saw you beat me to the punch 😂

I'll put together the test for #6504 now

This is a regression test for part of pytest-dev#6492, testing one of the fixes in pytest-dev#6551.

SalmonMode · 2020-01-24T21:06:25Z

@blueyed added the test case and also cleaned up the logic a bit by putting it into its own function so the code reads a little easier.

timc13 · 2020-02-03T05:56:14Z

Not sure if this is related, but I'm seeing that indirect fixtures are not getting cleaned up when being stacked.

e.g.

import pytest


@pytest.fixture
def x(): pass


@pytest.fixture
def y(): pass


@pytest.mark.parametrize("x", [0, 1], indirect=['x'])
@pytest.mark.parametrize("y", [2, 3], indirect=['y'])
def test_foo(x, y):
    pass

blueyed · 2020-02-03T06:03:56Z

@timc13 with this branch I assume?

timc13 · 2020-02-03T06:17:00Z

no, sorry - this is in 5.3.5. I was wondering if this PR is related. I can also create another ticket otherwise

blueyed · 2020-02-03T06:27:51Z

@timc13

I was wondering if this PR is related.

Test it? :)

I can also create another ticket otherwise

Yes, but information if this here helps would be useful there then also. (have not looked at the issue itself myself)

@pytest

Note: black cannot parse `return *active_fixture_argnames, *self.argnames` yet (fixed in master, psf/black#1121). Tested manually using: ```python @pytest.fixture(scope="session") def xdist_suffix(request): print("\nxdist_suffix") suffixes.append("xdist") @pytest.fixture(scope="session") def parallel_suffix(tox_suffix, xdist_suffix): pass def test_suffix(parallel_suffix): assert suffixes == ["tox", "xdist"] ``` When using a set there the order is not deterministic, i.e. the test is flaky.

…er finalizer to a dependee fixture once

… add ther finalizer to a dependee fixture once

Ref: pytest-dev/pytest#6551

SalmonMode · 2020-04-11T16:27:27Z

any update on this?

I also wanted to link this related fix for pytest-django pytest-dev/pytest-django#807 just for later reference

@timc13 were you able to run your tests using this branch to see if it solved your issue?

SalmonMode · 2020-05-12T12:01:36Z

@RonnyPfannschmidt @nicoddemus any thoughts on moving this forward? If I get the chance, I'll resolve the merge conflict later today (and I'm guessing that's the cause of the failed build). But I think it's a pretty solid fix at this point.

SHxKM · 2020-05-17T13:23:17Z

I'm pretty sure this is affecting me in Inconsistent results for a test when running in-suite. Is there at least a temporary workaround that once can use?

SalmonMode · 2020-05-17T13:30:15Z

@SHxKM would you be able to try this branch in combination with the branch for pytest-django here?

I believe both changes will provide the solution, because the original fix I made for this caused some issues to be revealed in pytest-django that line up with what you described in the issue you linked.

SHxKM · 2020-05-17T13:39:12Z

Edit: FWIW my issue had nothing to do with this.

@SalmonMode - Thanks for responding. So I ran:

pip install git+https://github.com/blueyed/pytest@fixture-stack
pip install git+https://github.com/SalmonMode/pytest-django@cnejame-fix-pytest-5-3-3-compatibility

I hope that's what you meant. The test still fails when running together with the whole suite :/

SalmonMode · 2020-05-17T13:44:21Z

Hmmm are you able to provide a minimal set of tests to recreate it?

nicoddemus · 2020-05-18T00:15:58Z

@RonnyPfannschmidt could you help reviewing this one as well?

RonnyPfannschmidt · 2020-05-18T05:44:48Z

at best by the weekend

blueyed · 2020-07-17T20:07:52Z

Closing as per blueyed#365 (comment).

SalmonMode · 2020-07-17T20:28:28Z

As mentioned here, I'll be remaking this branch under my repo again to continue the work.

blueyed · 2020-07-17T20:32:20Z

@SalmonMode Thanks!

blueyed added a commit to blueyed/pytest-django that referenced this pull request Jan 23, 2020

TEST: blueyed/pytest@fixture-stack

f0122e6

Ref: pytest-dev/pytest#6551

This was referenced Jan 23, 2020

TEST: blueyed/pytest@mm-fixture-stack pytest-dev/pytest-django#813

Closed

[WIP] Fix exponential recursion #6504

Closed

SalmonMode mentioned this pull request Jan 23, 2020

Restores the expected behavior of certain fixtures, while fixing a bug related to fixtures now tearing down properly pytest-dev/pytest-django#807

Open

blueyed mentioned this pull request Jan 24, 2020

Fixture tear-down order broken in 5.3.3 #6512

Open

blueyed changed the title ~~fixtures: register finalizers with all previous fixtures in the stack (take 2)~~ [WIP/RFC] fixtures: register finalizers with all previous fixtures in the stack (take 2) Jan 24, 2020

blueyed mentioned this pull request Jan 24, 2020

tests: add test_fixture_arg_ordering #6554

Merged

blueyed added a commit to blueyed/pytest that referenced this pull request Jan 24, 2020

tests: add test_fixture_arg_ordering

22a3b14

This is a regression test for part of pytest-dev#6492, testing one of the fixes in pytest-dev#6551.

nicoddemus pushed a commit to blueyed/pytest that referenced this pull request Jan 24, 2020

tests: add test_fixture_arg_ordering

192d3ad

This is a regression test for part of pytest-dev#6492, testing one of the fixes in pytest-dev#6551.

blueyed force-pushed the fixture-stack branch from 79ed509 to 6db4724 Compare January 25, 2020 18:29

This comment has been minimized.

Sign in to view

SalmonMode and others added 6 commits March 9, 2020 13:00

fixtures register finalizers with all fixtures before them in the stack

163b03c

avoid exponential recursion by checking if already added

cfee6e7

cleaned up, DAMPified, and added tests to ensure fixtures only add th…

a950653

…er finalizer to a dependee fixture once

add type annotations

b1e1b73

fixup! cleaned up, DAMPified, and added tests to ensure fixtures only…

4e07783

… add ther finalizer to a dependee fixture once

blueyed force-pushed the fixture-stack branch from f26c693 to 4e07783 Compare March 9, 2020 12:01

blueyed added a commit to blueyed/pytest-django that referenced this pull request Mar 31, 2020

TEST: blueyed/pytest@fixture-stack

0803b92

Ref: pytest-dev/pytest#6551

SalmonMode mentioned this pull request Apr 11, 2020

pytest_generate_tests() created fixture, together with pytest.mark.parametrize() created fixture (of different name) -- results in messed up fixture scope - and inappropriate re-instantiation of fixtures #6969

Open

SalmonMode mentioned this pull request May 17, 2020

Inconsistent results for a test when running in-suite pytest-dev/pytest-django#842

Closed

Merge branch 'master' into fixture-stack

9a87de6

blueyed mentioned this pull request Jun 30, 2020

Parametrized fixture regression in branch fixture-stack blueyed/pytest#365

Closed

blueyed closed this Jul 17, 2020

blueyed deleted the fixture-stack branch July 17, 2020 20:07

SalmonMode mentioned this pull request Jul 17, 2020

fixtures: register finalizers with all previous fixtures in the stack (take 3) #7511

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[WIP/RFC] fixtures: register finalizers with all previous fixtures in the stack (take 2) #6551

[WIP/RFC] fixtures: register finalizers with all previous fixtures in the stack (take 2) #6551

blueyed commented Jan 23, 2020 •

edited

blueyed commented Jan 24, 2020

SalmonMode commented Jan 24, 2020

SalmonMode commented Jan 24, 2020

This comment has been minimized.

This comment has been minimized.

timc13 commented Feb 3, 2020

blueyed commented Feb 3, 2020

timc13 commented Feb 3, 2020

blueyed commented Feb 3, 2020 •

edited

SalmonMode commented Apr 11, 2020 •

edited

SalmonMode commented May 12, 2020

SHxKM commented May 17, 2020

SalmonMode commented May 17, 2020

SHxKM commented May 17, 2020 •

edited

SalmonMode commented May 17, 2020

nicoddemus commented May 18, 2020

RonnyPfannschmidt commented May 18, 2020

blueyed commented Jul 17, 2020

SalmonMode commented Jul 17, 2020

blueyed commented Jul 17, 2020

[WIP/RFC] fixtures: register finalizers with all previous fixtures in the stack (take 2) #6551

[WIP/RFC] fixtures: register finalizers with all previous fixtures in the stack (take 2) #6551

Conversation

blueyed commented Jan 23, 2020 • edited

blueyed commented Jan 24, 2020

SalmonMode commented Jan 24, 2020

SalmonMode commented Jan 24, 2020

This comment has been minimized.

This comment has been minimized.

timc13 commented Feb 3, 2020

blueyed commented Feb 3, 2020

timc13 commented Feb 3, 2020

blueyed commented Feb 3, 2020 • edited

SalmonMode commented Apr 11, 2020 • edited

SalmonMode commented May 12, 2020

SHxKM commented May 17, 2020

SalmonMode commented May 17, 2020

SHxKM commented May 17, 2020 • edited

SalmonMode commented May 17, 2020

nicoddemus commented May 18, 2020

RonnyPfannschmidt commented May 18, 2020

blueyed commented Jul 17, 2020

SalmonMode commented Jul 17, 2020

blueyed commented Jul 17, 2020

blueyed commented Jan 23, 2020 •

edited

blueyed commented Feb 3, 2020 •

edited

SalmonMode commented Apr 11, 2020 •

edited

SHxKM commented May 17, 2020 •

edited