Re: [PATCH v2 7/7] var(win32): do report the GIT_SHELL_PATH that is actually used

From: Junio C Hamano <gitster@pobox.com>
Date: Jul 12, 2024, 15:35 UTC
Message-ID: <xmqqikxaoc7s.fsf@gitster.g>
In-Reply-To: <8bfd23cfa00c351ffdcc25bd29f3b84089544a56.1720739496.git.gitgitgadget@gmail.com>

"Johannes Schindelin via GitGitGadget" <gitgitgadget@gmail.com> writes:

Show 7 quoted lines

> From: Johannes Schindelin <johannes.schindelin@gmx.de>
>
> On Windows, Unix-like paths like `/bin/sh` make very little sense. In
> the best case, they simply don't work, in the worst case they are
> misinterpreted as absolute paths that are relative to the drive
> associated with the current directory.
> To that end, ...

This does not quite explain why a hardcoded full path does not make sense, though. If you do not want "relative to the current drive", you can even hardcode the path including the drive letter and now it no longer falls into that "worst case" because there is no way to misinterpret it as such. I think the real reason is that the port's "relocatable" nature does not mix well with the "let's make sure we know what we use exactly by deciding the paths to various tools at compile time" mentality. So if I were rerolling this I would say

	Git ported to Windows expect to be installed at user-chosen
	places, which means hardcoded paths to tools decided at
	compile time are bad fit.

or something.

But it does not matter if the stated reason is to use the "find 'sh' that is on PATH" approach was taken tells the whole story or not. The important thing is that we do ...

Show 6 quoted lines

> Git does not actually use the path `/bin/sh` that is
> recorded e.g. when `run_command()` is called with a Unix shell
> command-line. Instead, as of 776297548e (Do not use SHELL_PATH from
> build system in prepare_shell_cmd on Windows, 2012-04-17), it
> re-interprets `/bin/sh` as "look up `sh` on the `PATH` and use the
> result instead".

... this. And with the design constraints that it has to be installable anywhere and cannot rely on compile-time determined hardcoded paths, the above design decision is a very valid one.

> This is the logic users expect to be followed when running `git var
> GIT_SHELL_PATH`.

And matching `git var GIT_SHELL_PATH` to what the code path does make sense.

Show 19 quoted lines

> diff --git a/builtin/var.c b/builtin/var.c
> index 5dc384810c0..e30ff45be1c 100644
> --- a/builtin/var.c
> +++ b/builtin/var.c
> @@ -12,6 +12,7 @@
>  #include "refs.h"
>  #include "path.h"
>  #include "strbuf.h"
> +#include "run-command.h"
>  
>  static const char var_usage[] = "git var (-l | <variable>)";
>  
> @@ -51,7 +52,7 @@ static char *default_branch(int ident_flag UNUSED)
>  
>  static char *shell_path(int ident_flag UNUSED)
>  {
> -	return xstrdup(SHELL_PATH);
> +	return git_shell_path();
>  }

Simple and clear.  Nicely concluded.

Thanks.  Will queue.