Tests failing in random order (again)

[IndrajeetPatil]

@etiennebacher Do you want to have a look at it?

[etiennebacher]

I use the same script and the same seed but I don't have the same file order. For example, here:

• the first is test-ivreg.R but mine is test-ivreg_AER.R
• then it's the same until test-format_table.R which for me is test-format_table_ci.R
• etc.

So I can't reproduce all of these failures

[IndrajeetPatil]

Hmm, maybe because you are on Windows, and we run the check on Ubuntu for our CI/CD. On macOS, I can reproduce two failures.

I have changed the workflow so that the tests run on Windows. Let's see if we still see any test failures.

[IndrajeetPatil]

Yeah, failures are also seen on Windows: https://github.com/easystats/insight/actions/runs/4697252164/jobs/8328125120

Note that the seed is going to be different for each run.

[etiennebacher]

Hmm, maybe because you are on Windows, and we run the check on Ubuntu for our CI/CD. On macOS, I can reproduce two failures.

The problem is not due to this, it's that before running the tests I can't get the same order as you even if I use exactly the same seed (and I don't know why)

[strengejacke]

I see this warning:

Package logistf is loaded and breaks mmrm::mmrm() (1)

[etiennebacher]

It's not a warning, it's just a skip message. It's because of namespace conflict, cf https://stackoverflow.com/questions/72533745/loading-logistf-breaks-mcmcglmm/72536046

I made a PR upstream but no answer so far georgheinze/logistf#54