Re: [PATCH 3/4] string-list: change "string_list_find_insert_index" return type to "size_t"

> Patrick Steinhardt <ps@pks.im> writes:
> 
> >> @@ -266,7 +265,7 @@ static struct string_list_item *lookup_prefix(struct string_list *map,
> >>  	 * overlong key would be inserted, which must come after the
> >>  	 * real location of the key if one exists.
> >>  	 */
> >> -	while (0 <= --i && i < map->nr) {
> >> +	while (i-- > 0 && i < map->nr) {
> >
> > This could simply be `while (i-- && i < map->nr)`.
> 
> Yes, especially if the reason why we avoid "not negative" aka "0 <="
> is because the counter is now unsigned, yours is much more intuitive
> way to say "as long as i is not yet zero".  Alternatively you could
> say "while (i-- != 0 && ...", but not comparing with 0 is more
> customary.
> 
> Better yet, shouldn't we stay away from "i", if the point of the
> change is to make it unsigned, as "i" has a strong connotation with
> "int, the platform natural signed integer type"?