Hey all, As I understand it, the `on_exit/2` callback is *not* guaranteed to complete before a test process exits *if the test has a failure*. For example, consider this flow:
``` 1. Boot a process tree (Supervisor with child processes) 2. Register an `on_exit/2` callback to stop the Supervisor 3. Make an assertion -- the assertion fails ``` Here, the assertion failure will crash the test process before `on_exit/2` runs. Assuming I'm understanding this correctly, let me share a very common pain point for our team. A test will fail, and we'll see an output like this: ``` ** (Mox.UnexpectedCallError) no expectation defined for MyMock.some_func/0 1) some failing test match (=) failed :ok = Module.some_operation() ``` The issue is that we get error output *that was actually a result of the test pid being torn down before the process tree*. So, it's a red herring. There was *another* bug that caused the assertion to fail. In this case, the Mox.UnexpectedCallError just happened because the test_pid went DOWN while the process tree was still running, and they called a mock registered to that test_pid and failed. However, in certain situations, it's super hard to know if the error output is signal or noise. You can imagine that in the place of a Mox error, we have the same situation with database connections, no_procs when we try to send signals back to a dead test_pid, etc. We think it would be ideal to be able to register a callback that will always run before a test_pid exits, regardless of if the test ran successfully or not. What do you think? Or what other strategies might we employ? Thanks, Anthony -- You received this message because you are subscribed to the Google Groups "elixir-lang-core" group. To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-core+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/52c85fb1-6a4e-4646-910f-aeb95ec90efdn%40googlegroups.com.