Re: Running 'git worktree add' in 'pre-commit' hook

From: Jeff King <peff@peff.net>
Date: May 16, 2019, 22:17 UTC
Message-ID: <20190516221702.GA11784@sigill.intra.peff.net>
In-Reply-To: <CACsJy8DnkjuZD-9pbhAsFo16jHKt8U831LLxb3-nCQP5_FOmtA@mail.gmail.com>

On Thu, May 16, 2019 at 06:25:24PM +0700, Duy Nguyen wrote:

Show 13 quoted lines

> > Is it forbidden to call this command from a hook?
> 
> pre-commit hook sets GIT_INDEX_FILE to this "index.lock" so you have
> the latest index content (which is not the same as from
> $GIT_DIR/index). This variable will interfere with any commands that
> work on a different worktree.
> 
> So you probably can still make it work by backing up $GIT_INDEX_FILE
> (in case you need it), then unset it before you use "git worktree" (or
> cd to it if you keep a permanent separate worktree for pre-commit
> activities). To make sure you don't have similar problems, you
> probably should do "env | grep GIT" from the hook and see if any other
> variables are set.

If you're entering another repo from a hook, you're supposed to use:

  unset $(git rev-parse --local-env-vars)

I wondered if we'd need another similar mechanism for entering the worktree of another repo, that would maybe clear fewer variables. But I think it's actually the same: we really want to clear everything and let our "cd" pick up the new repository path.

The case of actually _adding_ a new work tree (before we enter it) is weirder, though. We definitely want to stay in the same repository, and clearing all of that would not make sense. I do wonder if worktree-add should be handling GIT_INDEX_FILE (ignoring it when we want to be dealing with the index of the new worktree we added, and handling any relative fixups if we chdir inside the worktree code).

-Peff