Re: [PATCH] pack-objects: use strcspn(3) in name_cmp_len()

["Ævar Arnfjörð Bjarmason" <avarab@gmail.com>]

On Sun, Feb 05 2023, René Scharfe wrote:

> -static int name_cmp_len(const char *name)
> +static size_t name_cmp_len(const char *name)
>  {
> -	int i;
> -	for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
> -		;
> -	return i;
> +	return strcspn(name, "\n/");
>  }

I wonder if this name_cmp_len() is worth keeping at all. If all we're
doing is wrapping strcspn() (which b.t.w, seem to be less "open-coding"
and just that it wasn't known to the original author in 5d4a6003354
(Make git-pack-objects a builtin, 2006-08-03)), then just inlining that
in the two name_cmp_len() invocations would be better, or maybe:

	strcspn(..., object_reject);

Where we previously declared "object_reject" (or same better name for
such a variable) to be "\n/".

>  static void add_pbase_object(struct tree_desc *tree,
>  			     const char *name,
> -			     int cmplen,
> +			     size_t cmplen,
>  			     const char *fullname)
>  {
>  	struct name_entry entry;
>  	int cmp;
>
>  	while (tree_entry(tree,&entry)) {
>  		if (S_ISGITLINK(entry.mode))
>  			continue;
>  		cmp = tree_entry_len(&entry) != cmplen ? 1 :
>  		      memcmp(name, entry.path, cmplen);

The return of tree_entry_len() is "int", so that's a new implicit cast,
but that seems OK in this case (and the "namelen" is another thing we
should convert to "size_t").