Parametrized fixture regression in branch fixture-stack

[oakkitten]

as reported here and confirmed by my tests, branch fixture-stack runs a parametrized fixture a suboptimal amount of times.

running this example, output using current upstream master (5.4.1.dev422+gcf640a9b6, also 3.10.1 and some versions in between):

test_cats_dogs.py dog charlie
cat tom
.sssdog buddy
ss.scat simba
sss.dog charlie
s.ss

output from fixtue-stack:

test_cats_dogs.py dog charlie
cat tom
.sssdog buddy
cat tom
ss.scat simba
sss.dog charlie
cat simba
s.ss

[blueyed]

Thanks for the report. Should go into the discussion on/of the PR itself though: pytest-dev#6551.

(FYI: I do not plan to finish/work on this myself anytime soon in the pytest-dev/pytest project itself.)

[SalmonMode]

@blueyed as discussed on pytest-dev#7053, that is not a problem with the fixture-stack branch. That was a misunderstanding of the description of how pytest "optimizes" the fixtures it runs. I will remake the branch and take over the work again though.