jujutsu test failure on Nix install

[mitchellh]

Steps to Reproduce the Problem

I've added the jujutsu flake and overlay to my flake as described.

This previously has worked for many versions, it only recently started failing. I'm sorry I haven't bisected it.

Actual Behavior

error: builder for '/nix/store/q0g3dzx7qdx2vz6rw9qx8sc6b7prfkap-jujutsu-unstable-0a73245.drv' failed with exit code 100;
       last 25 log lines:
       >  right: [CommitId("0000000000000000000000000000000000000000")]
       > stack backtrace:
       >    0: rust_begin_unwind
       >              at /rustc/e92993dbb43f0a5d17fe56e2d82f90435d6521c8/library/std/src/panicking.rs:665:5
       >    1: core::panicking::panic_fmt
       >              at /rustc/e92993dbb43f0a5d17fe56e2d82f90435d6521c8/library/core/src/panicking.rs:74:14
       >    2: core::panicking::assert_failed_inner
       >    3: core::panicking::assert_failed
       >    4: core::ops::function::FnOnce::call_once
       >    5: core::ops::function::FnOnce::call_once
       >              at /rustc/e92993dbb43f0a5d17fe56e2d82f90435d6521c8/library/core/src/ops/function.rs:250:5
       > note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
       >
       >    Canceling due to test failure: 5 tests still running;1mpassed, 0 skipped
       >         PASS [   0.016s] jj-lib::runner test_gpg::invalid_signaturefailed, 0 skipped
       >         PASS [   0.020s] jj-lib::runner test_gpg::unknown_keyiled, 0 skipped
       >         PASS [   0.018s] jj-lib::runner test_id_prefix::test_id_prefix 0 skipped
       >         PASS [   0.012s] jj-lib::runner test_id_prefix::test_id_prefix_divergent skipped
       >         PASS [   0.024s] jj-lib::runner test_gpg::gpg_signing_roundtrip0 skipped
       >         PASS [   0.025s] jj-lib::runner test_gpg::gpg_signing_roundtrip_explicit_key3;1mskipped
       > ------------[0m [ 00:00:08] [] 1189/1421: 1 running, 1188 passed, 1 failed, 0 skipped
       >      Summary [   8.474s] 1190/1421 tests run: 1189 passed, 1 failed, 0 skipped
       >         FAIL [   0.012s] jj-lib::runner test_git::test_shallow_commits_lack_parents
       > warning: 231/1421 tests were not run due to test failure (run with --no-fail-fast to run all tests)
       > error: test run failed90/1421: 0 running, 1189 passed, 1 failed, 0 skipped
       For full logs, run 'nix log /nix/store/q0g3dzx7qdx2vz6rw9qx8sc6b7prfkap-jujutsu-unstable-0a73245.drv'.

Specifications

• Platform: Linux dev 6.11.5 Fix readme typo #1-NixOS SMP Tue Oct 22 13:51:37 UTC 2024 aarch64 GNU/Linux
• Version: 7ba8d73

[PhilipMetzger]

The shallow commit test is flaky, if I recall correctly. So if there's a way to rebuild in Nix, you should be able to get a functional binary. cc @thoughtpolice @emilazy for more Nix expertise.

[mitchellh]

I've tried a rebuild 3 or 4 times and it has failed consistently. It's hard to keep trying because it takes a very long time for jujutsu to rebuild (since the flake rebuilds every dependency each time). If I go back to the prior version, it builds/passes the first time and I've never seen the test fail, so perhaps something was introduced to make it extra flaky?

[thoughtpolice]

I reported this last week in Discord but I haven't had much hacking time in the past week. It fails consistently for me on WSL2. We should probably just disable this test for now.

See also #3537. I'd be interested to know if it was a similar coarse-grained mtime issue, or something (running the testsuite on an ext3 loopback device should be enough to do that.)

[PhilipMetzger]

This should be fixed for the latest trunk, see #4839 (comment). @mitchellh can you rebuild the flake again?

[mitchellh]

Thanks, I'm out for today but should be able to give it a test run tonight.

[PhilipMetzger]

feel free to close the issue if succeeds.

[mitchellh]

Fixed. Thanks.