Re: [PATCH v3 5/7] refs: add safe_create_reflog function

From: Junio C Hamano <gitster@pobox.com>
Date: Jun 26, 2015, 22:12 UTC
Message-ID: <xmqqlhf6rudx.fsf@gitster.dls.corp.google.com>
In-Reply-To: <1435278548-3790-5-git-send-email-dturner@twopensource.com>

David Turner <dturner@twopensource.com> writes:

Show 8 quoted lines

> Make log_ref_setup private, and add public safe_create_reflog which
> calls log_ref_setup.
>
> In a moment, we will use safe_create_reflog to add reflog creation
> commands to git-reflog.
>
> Signed-off-by: David Turner <dturner@twopensource.com>
> ---

Show 8 quoted lines

> @@ -629,9 +628,8 @@ static void update_refs_for_switch(const struct checkout_opts *opts,
>  				ref_name = mkpath("refs/heads/%s", opts->new_orphan_branch);
>  				temp = log_all_ref_updates;
>  				log_all_ref_updates = 1;
> -				ret = log_ref_setup(ref_name, &log_file, &err);
> +				ret = safe_create_reflog(ref_name, &err, 0);
>  				log_all_ref_updates = temp;
> -				strbuf_release(&log_file);

What the result of applying this patch does around here is incoherent, isn't it? I do understand and agree that adding the "force-create" argument to safe_create_reflog() is a good idea, but then I do not think you need to temporarily set log-all-ref-updates while calling it (and revert it afterwards). Not having to do that is the whole point of "force-create", isn't it?

Show 11 quoted lines

> diff --git a/refs.c b/refs.c
> index dff91cf..7519dac 100644
> --- a/refs.c
> +++ b/refs.c
> @@ -3123,11 +3123,12 @@ static int should_autocreate_reflog(const char *refname)
>  	return starts_with(refname, "refs/heads/") ||
>  		starts_with(refname, "refs/remotes/") ||
>  		starts_with(refname, "refs/notes/") ||
> +		!strcmp(refname, "refs/stash") ||
>  		!strcmp(refname, "HEAD");
>  }

This change is *not* part of creating safe-create-reflog.

It may be part of "allowing us to use safe_create_reflog() to manage stash", but I'd imagine that a new "reflog create" command should ignore log-all-ref-updates settings (after all, the end user is explicitly telling us to create one by issuing a command), so I doubt it matters if 'stash' is on the auto-create list.

Show 13 quoted lines

>  /* This function will fill in *err and return -1 on failure */
> -int log_ref_setup(const char *refname, struct strbuf *sb_logfile, struct strbuf *err)
> +static int log_ref_setup(const char *refname, struct strbuf *sb_logfile, struct strbuf *err, int force_create)
>  {
>  	int logfd, oflags = O_APPEND | O_WRONLY;
>  	char *logfile;
> @@ -3136,7 +3137,8 @@ int log_ref_setup(const char *refname, struct strbuf *sb_logfile, struct strbuf
>  	logfile = sb_logfile->buf;
>  	/* make sure the rest of the function can't change "logfile" */
>  	sb_logfile = NULL;
> -	if (log_all_ref_updates && should_autocreate_reflog(refname)) {
> +	if (log_all_ref_updates &&
> +	    (force_create || should_autocreate_reflog(refname))) {

With this patch, force_create is used as "we are throwing you an unusual name that do not automatically get a reflog when log-all is set, and we are overriding that selection-by-name aspect of auto creation" but does not defeat log-all settings. Is that intended?

If somebody asks safe_create_reflog() to create a reflog and passes force_create, shouldn't we create that reflog regardless of log_all_ref_updates settings? That is

	if (force_create || should_autocreate_reflog(refname)) {
        	... create it ...

would be more natural to read here, and you would get that for free if 4/7 checked log_all_ref_updates in should_autocreate_reflog().