Jim Kane

Hat tip to @aaronsakowski who did the actual work. I read and followed his repo changes to create this PR.

> Thanks for putting together the PR @fastjames! Glad to help! I have a branch for https://github.com/danschultzer/phoenix_oauth2_provider that should provide the same updates, ready to PR once this is merged.

> @fastjames Hopefully this is helpful - I noticed the migrations rely on the attrs defined in the modules so this PR will cause the migrations to be generated with...

I was able to access an older Intel-based mac (mac mini running macOS Catalina 10.15.7), so I installed elixir/erlang and ran the same ex-gpgme tests there. The tests passed with...

I was able to extract some information about my running `mix` process using Activity Monitor. Below is the sampling output, with a call graph that appears to indicate a panic...

@evnu thanks very much for the pointer to that. By scheduling this function on the DirtyIO scheduler, I was able to get the test to time out instead of hanging...

I ran the same test suite under OTP 22 and got the following output: ``` ❯ GPGME_DEBUG=9 mix test Excluding tags: [:manual_pinentry] from_protocol: after arg transform protocol is CMS rust...

Today I was able to use dtruss to track the activity of gpgconf as it was called, and according to the output it appears to have run at least to...

Today I reran the previous dtruss test under OTP22. The final call from gpgconf is identical to what I saw under OTP 23, which confirms that `gpgconf --list-dirs` is completing...

I repeated the dtruss / mix process under OTP 22, and I observed that the reported wait PID was off by 1 from the PID of the gpgconf process. Therefore,...