Re: [PATCH] remote: prefer exact matches when using refspecs

From: Jonathan Tan <jonathantanmy@google.com>
Date: Jul 31, 2018, 23:33 UTC
Message-ID: <20180731233332.187328-1-jonathantanmy@google.com>
In-Reply-To: <xmqq1sbj9h08.fsf@gitster-ct.c.googlers.com>

> That is, something like this, perhaps.  The resulting behaviour
> should match how "git rev-parse X" would give precedence to tag X
> over branch X by going this route.  What do you think?

[snip]

Show 19 quoted lines

>  static const struct ref *find_ref_by_name_abbrev(const struct ref *refs, const char *name)
>  {
>  	const struct ref *ref;
> +	const struct ref *best_match = NULL;
> +	int best_score = -1;
> +
>  	for (ref = refs; ref; ref = ref->next) {
> -		if (refname_match(name, ref->name))
> -			return ref;
> +		int score = refname_match(name, ref->name);
> +
> +		if ((score && (best_score < 0 || score < best_score))) {
> +			best_match = ref;
> +			best_score = score;
> +		}
>  	}
> -	return NULL;
> +	return best_match;
>  }

This looks good to me. I've checked that refname_match (and branch_merge_matches(), which returns the result of refname_match() directly) is only used in "if" contexts, so making it return a value other than 1 is fine.

I would initialize best_score to INT_MAX to avoid needing the "best_score < 0" comparison, but don't feel strongly about it.