On OSX run:
bats test
Tests should pass as they were before.
Many failing tests are encountered. And some tests hang indefinitely. An example error I captured from a single test:
/Users/trevorbrown/.asdf/bin/asdf: line 82: /var/folders/7m/9bq62tmj60zf14d788w0lk8c0000gp/T/asdf.XXXX.ed4MwuLg/home/.asdf/lib/commands/command-help: No such file or directory
OS: 10.15.1
asdf version: v0.7.6-bde7ae5
@vic do you know what is going on? I've been running bats test for years and I'm not sure why I'm seeing so many failures now.
The asdf update is a noop for non-git repos test hangs indefinitely.
I am having the same issue, which makes debugging #633 challenging. Manually running asdf with my changes seems to work as expected.
On tag v0.7.5 I only get one failure:
✗ get_config returns default when config file does not exist
(in test file test/get_asdf_config_value.bats, line 29)
`[ "$result" = "no" ]' failed
On tag v0.7.6 I get 65 failures before it reaches the asdf update is a noop for non-git repos test and it hangs indefinitely.
I get the exact same failures and hanging test on my Ubuntu laptop as well.
Try using https://github.com/bats-core/bats-core (just like our CI is doing) and not https://github.com/sstephenson/bats, the later has not been updated in years.
@vic I am using bats-core version 1.1.0. I don't think the bats version is the issue.
Do test pass for you @vic ? What OS are you on?
Yep, all green here. 212 test, 0 failures.
Also green in CI macos catalina / ubuntu.
https://github.com/asdf-vm/asdf/runs/375715344#step:7:219
I'm using Mac OS Catalina.
bats-core 1.2.0-dev (commit 87b16eb8ec381beca7f05d028e4ad7cf3f647a0f)
git --version
git version 2.21.0 (Apple Git-122)
I ran git bisect with the failing tests (git bisect run bats test/current_command.bats) and got:
e968fe6142d121d1da11aeada21b0338b62b9b16 is the first bad commit
commit e968fe6142d121d1da11aeada21b0338b62b9b16
Author: Victor Hugo Borja <[email protected]>
Date: Wed Nov 27 02:24:29 2019 -0600
Subcommand-aware command line interface.
This is a very large commit, and it's not immediately clear what the problem is, but I am continuing to dig.
CI seems to be failing as well, and has been since November. Any idea what's going on there? Is that a separate issue or the same one?
@lahwran The only failing check is the Appveyor one, it's still a work in progress (we have never been successful in running tests under Windows). Maybe we should remove the appveyor check until it's stable.

I found the problem! Because I'm running asdf on my computer ASDF_DIR was already set and was interfering with the tests. By unsetting the variable all tests pass:
unset ASDF_DIR
bats test
The asdf update is a noop for non-git repos test is still hanging for me. I'm guessing it is caused by a similar issue. Will investigate later.
I created https://github.com/asdf-vm/asdf/pull/644 to fix that hanging test.
Most helpful comment
I found the problem! Because I'm running asdf on my computer
ASDF_DIRwas already set and was interfering with the tests. By unsetting the variable all tests pass:The
asdf update is a noop for non-git repostest is still hanging for me. I'm guessing it is caused by a similar issue. Will investigate later.