Keeping it as a 32-bit value means the min size calculation overflows or gets truncated which can lead to issues with large packfiles.
