Re: [PATCH 4/4] drop git_exec_path() from non-Git commands' PATH

[Patrick Steinhardt <ps@pks.im>]

On Wed, May 21, 2025 at 02:07:25PM +0100, Phillip Wood wrote:
> On 21/05/2025 09:27, Patrick Steinhardt wrote:
> > I agree with what Junio mentioned in a parallel thread, especially
> > around Git hooks. The expectation there is that those may execute other
> > Git commands, and that should typically be using the same execution
> > environment as the original Git command that has been invoking the hook.
> > So refining this patch so that the mechanism is opt-in probably makes
> > sense.
> > 
> >      A slight tangent: I wonder whether it is even required nowadays to
> >      adapt PATH at all anymore. As far as I understand this was a
> >      requirement back when people still executed dashed binaries
> >      directly. But nowadays scripts don't really do that anymore, but
> >      instead use the git binary. And that one doesn't need PATH to be
> >      adapted at all, as it knows to listen to GIT_EXEC_PATH and its
> >      built-in path anyway.
> 
> But don't we still need to change PATH so that hooks, shell aliases, git
> rebase --exec, git bisect run, etc. still run the same git executable that
> started them? For example "/usr/bin/git -c alias.g=!git g --version" should
> report the version of /usr/bin/git, not ~/.local/bin/git which comes first
> in my PATH if git doesn't change it.

There's two parts to this: PATH and GIT_EXEC_PATH. We do have to adjust
PATH indeed to contain the location of the 'git' executable. But we also
add GIT_EXEC_PATH to it, which I'm less sure whether it's actually
needed.

Patrick