Hi @remcohaszing,

Thanks :)

Did you create hooks with husky add (or husky set)? It should create them with husky.sh being sourced at the beginning of the file.

Good suggestion for the FAQ.

No, I created .husky/pre-commit with the following contents:

#!/usr/bin/env sh

yarn lint-staged

And of course I added the following to package.json:

  // …
  "scripts": {
    // …
    "prepare": "husky install"
  },
  // …

This just works.

husky set ... doesn't seem to be documented anywhere on https://typicode.github.io/husky/#/

@typicode, wondering if husky should use #!/usr/bin/env sh instead of #!/bin/sh in all it's scripts.

Generally speaking, I'd recommend keeping husky.sh. It contains code that can be useful for GUI users cloning the project and may have some additional features in future versions of husky.

@thasmo I picked the shebang used by Git hooks samples (in .git/hooks). I'm not against env sh but what would be good reasons to change it?

Using #!/usr/bin/env some-shell allows some-shell to be detected from the environment, whereas specifying an exact path in the shebang forces the user to install the shell to that specific location. Some practival examples:

• #!/usr/bin/env node works with nvm.
• #!/usr/bin/env python works with Python virtual environments.
• #!/usr/bin/env bash allows users to install a specific version of bash. I.e. I recall system bash was outdated on MacOS at some point.

For husky.sh the shebang is unused. It’s supposed to be sourced from a script, not to be called as a script directly.

All of my questions have been answered. Thanks!

Why is husky.sh added to the .gitignore file? This means anyone who checks out the code in my repo can still commit without the hooks running. The hooks wont run until the user runs npx husky install to generate the _ folder that's missing from the repo since it's being ignored?

adding a prepare script to package.json solved it for me.

{ "scripts": {"prepare": "husky install"}}

Every time I checkout from my branch to master, the husky.sh file disappears. Any reason why> Why can it not be committed with the repo?

@joetidee I don’t know why you’d want that, but you can use git add -f to ignore the .gitignore file.

git add -f .husky/_/husky.sh

@joetidee I don’t know why you’d want that, but you can use git add -f to ignore the .gitignore file.

git add -f .husky/_/husky.sh

I'm not sure why I wouldn't want that :) - could you elaborate on why it shouldn't be included?

I'm not sure why I wouldn't want that :) - could you elaborate on why it shouldn't be included?

It’s autogenerated by husky. Generally I think committing autogenerated files is a bad idea. Husky creates a .gitignore file for it for a reason.

ok, so why doesn't it autogenerate when I checkout master? (I have the "prepare": "husky install" in my package.json)
p.s. I have husky 6.0.0 installed.

The prepare script in package.json run when you run npm install (or yarn), not when you switch branches.

Right, so if it disappears when switching branches, that is a pain in the backside :)

Switching branches doesn’t delete files that haven’t been committed. Maybe you have committed it in one of the branches your working on?

@husky team - not including this shell script and forcing new repo users to do the init overrides the existing repo's .husky/pre-commit file.

Not very cash money....