But for data we can use the seed files and associated patterns, right? Since it's just text in those files, they're trivially checked into source control, diffed, distributed etc.