Conversation
It's somewhat confusing, but it seems sometimes `git` uses a shell, but other times it doesn't. And that is differentiate by the type system. It would still have to be (re)validated at some point if this is actually the case, but I guess there was a reason for this in the first place.
|
Thanks a lot for contributing! I think this is exactly how it would be done, and seeing Please note that when using the value of Let me fix that… |
That way it's possible to obtain an executable, program or script from a key in the configuration that is in a trusted section of the configuration. This goes along with a new `command` feature that brings in the `command` module at the top level to be able to execute such commands.
|
Now there also is a I hope that will help. |
Happy to adjust based on feedback, I wanted to add a feature to danobi/prr#36 where we would use
$EDITORas a fallback, and usecore.editoras the primary source, to match expectations withgititself, but the field isn't (yet) supported here, so this PR adds thecore.editorfield, I hope.There is a test failing, but it seems like an unrelated failure, and it might be my local setup interfering, as it's about
file::init::comfort::from_git_dir, so I want to open the PR to let the CI take a shot at it.Looking forward to your feedback, feel free to edit and adjust as you see fit.