Feature request: do not abort if single test binary aborts

[sebschmi]

Motivation

I have a workspace with a few crates, one of which uses some python dependencies that fail to compile on some machines. However, this prevents cargo nextest run from working at all:

$ cargo nextest run
    Finished `test` profile [unoptimized + debuginfo] target(s) in 0.08s
error: creating test list failed

Caused by:
  for `py_lib_tsalign`, command `/home/sebschmi/git/template-switch-aligner/target/debug/deps/tsalign-b8149d0ba639ceef --list --format terse` exited with exit code 127
  --- stdout:

  --- stderr:
  /home/sebschmi/git/template-switch-aligner/target/debug/deps/tsalign-b8149d0ba639ceef: error while loading shared libraries: libpython3.12.so.1.0: cannot open shared object file: No such file or directory

  ---

Proposal

Instead of aborting, collect tests from binaries that work only. If there is a broken binary, still print the error message somehow, such that the users knows that not all tests are ran as expected.

Alternatives

No response

Additional context

No response

[sunshowers]

Thanks for the request. I don't think nextest should support this kind of feature because it will tend to hide bugs. I'd recommend using a filterset on impacted machines.

[sebschmi]

It will not hide bugs if nextest still outputs the corresponding error messages, and exits with a non-zero exit code in the end.