Sleep Sort in X86-64 Locks Up

[rzuckerm]

This seems to happen pretty randomly. I can reproduce the issue locally maybe about half the time with this:

poetry run glotter test -l x86-64

Here is an example of the failure. From the failure, it appears that the last entry is not being output. I'm moving archive/x/x86-64/sleep-sort.asm to needs-attention/archive/x/x86-64/sleep-sort.asm.

Copying the original author: @RMarx1456

[RMarx1456]

It is definitely something to do with the cpu timing differences between the machines and environments. I will work on it when I can again. Thank you.

[RMarx1456]

When I originally made the pull request, it passed all of the test cases, but when mulitiple threads are involved, there's more of a "randomness" to it, so that is likely why it wasn't passing the test cases for that instance.

[RMarx1456]

My approach might be to spinlock the main thread until the child is done preparing itself (instead of waiting), making the input iterations more strict (whitelisting input characters), and more gracefully terminating the threads (have main send a SIGTERM to the threads).

[RMarx1456]

        actual = sleep_sort.run(params=in_params)
        actual = actual.replace("[", "").replace("]", "").strip()
>       assert actual == expected
E       AssertionError: assert '1, 2, 3, 4, 4,' == '1, 2, 3, 4, 4, 5'
E         - 1, 2, 3, 4, 4, 5
E         ?               --
E         + 1, 2, 3, 4, 4,

test/generated/test_sleep_sort.py:37: AssertionError
=========================== short test summary info ============================
FAILED test/generated/test_sleep_sort.py::test_sleep_sort_valid[x86-64/sleep-sort.asm-sample input: with duplicate] - AssertionError: assert '1, 2, 3, 4, 4,' == '1, 2, 3, 4, 4, 5'
  - 1, 2, 3, 4, 4, 5
  ?               --
  + 1, 2, 3, 4, 4,
================== 1 failed, 289 passed in 3731.85s (1:02:11) ==================

It's missing the five, so that tells me it's an issue with how the last thread is getting handled, which is the same issue I had on my own machine while first writing this.