Timing issue after using `set_mox_global`

[Sinc63]

I upgraded Mox from 1.1.0 to 1.2.0 and now have an intermittent problem.

I have two test files in a directory, the first of which sets Mox to global mode, and the second of which does not.

I've never had problems with these files before, but since upgrading to 1.2.0 the second file will sometimes now fail multiple tests with an error like:

     ** (ArgumentError) cannot add expectations/stubs to JitsiMeet.Token.JwtMock in the current process (#PID<0.1826.0>) because Mox is in global mode and the global process is #PID<0.1814.0>. Only the process that set Mox to global can set expectations/stubs in global mode

It's not consistent, but I would estimate it occurs on roughly 1/8 test runs. Looking at the code changes I would think this must have to do with NimbleOwnership.allow having a timeout. I didn't look farther than seeing that. I would think that the setting of the global mode should end when the test module exits, (i.e. from an on_exit function), not based on a timeout.

Erlang/OTP 26 [erts-14.2.5] [source] [64-bit] [smp:2:2] [ds:2:2:10] [async-threads:1] [jit:ns]

Elixir 1.14.5 (compiled with Erlang/OTP 26)

[whatyouhide]

Are both tests running with async: true?

[Sinc63]

Neither file is async.
If there is a diagnostic flag that I can enable to help debug this, or a place I could insert something to help diagnose this I can do that. Just tell me what and where.

[whatyouhide]

There isn't one really.

Looking at the code changes I would think this must have to do with NimbleOwnership.allow having a timeout

What makes you think that? The allow/5 call has a timeout of 30s, so I'd be surprised if it's that.

Do the PIDs mentioned in the error message correspond to any "known" PID that you can debug, like the test PID or something like that? This is going to be really hard to work through without a minimal reproducing example or being able to tinker with the situation 😔