Sometimes:

ERROR: Formula evaluation gave negative result for link 12

or link 2 is outputted, but the test requires

ERROR: Formula evaluation gave negative result for link 10

I assumed Windows was downloading a huge update during office hours again, or something else particular to my local test environment was causing the weird behaviour. But this test behaves very strangely in CI too. So I'm not sure if the test is flakey, overly reliant on the test environment, if it relies on a stochastic function within sDNA, or if there really is a bug (if so, how important is this part of the test? - do tests need to fail if error messages are wrong, failing for failing wrongly?).

I was able to get all the other tests to pass, by simply persevering, and rerunning any failed jobs. But not the Python 3.8 test, yet:

https://github.com/JamesParrott/sdna_plus/actions/runs/7907698164/job/21586911903

However it passed fine in Python 3.8.10 locally just now, so I doubt there's a Python 3.8 only bug:

(pyenv_3.8.10_pytest_numpy) C:\...\sdna_plus\sDNA\sdna_vs2008\tests\pytest>set DONT_TEST_ONE_LINK_SUBSYSTEMS_ORDER=1 & pytest
================================================= test session starts =================================================
platform win32 -- Python 3.8.10, pytest-8.0.0, pluggy-1.4.0
rootdir: C:\...\repos\sdna_plus\sDNA\sdna_vs2008\tests\pytest
collected 10 items

test_gathering_windows_command_line_tests.py ..........                                                          [100%]

================================================= 10 passed in 26.09s =================================================