Non-parity features that are large enough, or a radical enough departure from the status quo, that we should wait until Sweetie is mostly deprecated before prioritizing them.