Perhaps it's an attempt at optimization? A greedy RE will match the entire filename from that point on and then have to back track character by character to determine if there is or isn't a match. By making the RE non-greedy, matching filenames will be found faster.