git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [RFC/PATCH v2 2/3] run-command: make sure hooks have always GIT_DIR

From
Junio C Hamano <gitster@pobox.com>
Date
Apr 23, 2014, 20:37 UTC
Message-ID
<xmqqoazrye6w.fsf@gitster.dls.corp.google.com>
In-Reply-To
<1398282160-26151-3-git-send-email-felipe.contreras@gmail.com>
Felipe Contreras <felipe.contreras@gmail.com> writes:
> Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>

Why is this a good change? From a zero-line log message, I cannot even tell if this is trying to fix some problem, or trying to give new capabilities to hooks.

How does it prevent existing hook scripts from suddenly start misbehaving, where they do *not* expect to see an explicit GIT_DIR pointing at the original repository hook gets run in exported into their environment? For example, one of my post-receive hooks in a repository I push into records $cwd (which is the GIT_DIR of receiving repository), chdir's to another repository and then executes "git pull $cwd" from there, and that relies on the fact that being at the top-level of that other repository without GIT_DIR environment pointing at elsewhere but having .git directory in that top-level repository is sufficient to kick the auto-discovery of the repository that receives the "pull" in order to work correctly.

Show 55 quoted lines
> ---
>  run-command.c | 24 ++++++++++++++++++++++--
>  1 file changed, 22 insertions(+), 2 deletions(-)
>
> diff --git a/run-command.c b/run-command.c
> index 75abc47..8e188f6 100644
> --- a/run-command.c
> +++ b/run-command.c
> @@ -765,12 +765,29 @@ int run_hook_ve(const char *const *env, const char *name, va_list args)
>  	struct child_process hook;
>  	struct argv_array argv = ARGV_ARRAY_INIT;
>  	const char *p;
> -	int ret;
> +	const char **mod_env = NULL;
> +	int ret, i = 0;
> +	struct strbuf buf = STRBUF_INIT;
>  
>  	p = find_hook(name);
>  	if (!p)
>  		return 0;
>  
> +	if (!getenv(GIT_DIR_ENVIRONMENT)) {
> +		if (env)
> +			for (i = 0; env[i]; i++);
> +
> +		mod_env = xcalloc(i + 2, sizeof(*mod_env));
> +
> +		if (env)
> +			for (i = 0; env[i]; i++)
> +				mod_env[i] = env[i];
> +
> +		strbuf_addf(&buf, "GIT_DIR=%s", get_git_dir());
> +		mod_env[i++] = buf.buf;
> +		mod_env[i++] = NULL;
> +	}
> +
>  	argv_array_push(&argv, p);
>  
>  	while ((p = va_arg(args, const char *)))
> @@ -778,12 +795,15 @@ int run_hook_ve(const char *const *env, const char *name, va_list args)
>  
>  	memset(&hook, 0, sizeof(hook));
>  	hook.argv = argv.argv;
> -	hook.env = env;
> +	hook.env = mod_env ? mod_env : env;
>  	hook.no_stdin = 1;
>  	hook.stdout_to_stderr = 1;
>  
>  	ret = run_command(&hook);
>  	argv_array_clear(&argv);
> +	strbuf_release(&buf);
> +	free(mod_env);
> +
>  	return ret;
>  }
Previous: Felipe ContrerasNext: Felipe Contreras
Message 6 of 12 in “New 'update-branch' hook”
  1. 0/3 New 'update-branch' hookFelipe Contreras, Apr 23, 2014
  2. 1/3 sh-setup: export GIT_DIRFelipe Contreras, Apr 23, 2014
  3. Jeff KingApr 23, 2014
  4. Junio C HamanoApr 23, 2014
  5. 2/3 run-command: make sure hooks have always GIT_DIRFelipe Contreras, Apr 23, 2014
  6. Junio C HamanoApr 23, 2014
  7. Felipe ContrerasApr 23, 2014
  8. 3/3 Add 'update-branch' hookFelipe Contreras, Apr 23, 2014
  9. Junio C HamanoApr 23, 2014
  10. Felipe ContrerasApr 23, 2014
  11. Junio C HamanoApr 23, 2014
  12. Felipe ContrerasApr 23, 2014

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.