Closed dcarley closed 5 days ago
Thank you for the detailed bug report! Looking into it now.
Looks like this was fixed in 0.9.83, likely as part of the way outputs are collected being refactored (https://github.com/nextest-rs/nextest/commit/f1966f791c136d4d5ebd9e6c1c2b8d1ad0df3883). Could you update to the latest nextest (0.9.84 currently) and try again?
However it is concerning that I didn't realize this bug existed! I think it must have gotten introduced as a regression at some point.
https://github.com/nextest-rs/nextest/pull/1892 adds coverage for this situation.
Going to close this out as fixed in the latest version of nextest. Please reopen if you're still seeing it.
My bad, I should have checked the latest version.
Thanks for the quick response, additional test, and great project!
Description of the issue
Description: The output of a failing assertion is swallowed when the test also leaks a subprocess.
This makes it hard to debug the original cause of the failure, which might be that the clean-up was just never reached.
Steps to reproduce:
Run
cargo nextest run
with the following test that makes a failing assertion before the process is cleaned up:Expected outcome
Output from the
assert_eq
should be displayed along with the fact that it leaked.Fabricated output:
Actual result
We only get information that it failed and leaked, but not why it failed:
Nextest version