Schema is generated from meta model and various utils exist to drop/create tables/constraints/sequences.
OTOH, I have a metric assload of data that took a couple weeks to amass - this I've been keeping in a pg_dump generated file.
What about deployment - is this still the way to go for production? It kind of looks like pg_dump to archive file with pg_restore gives quite a bit more flexibility - but I am suspicious of binary formats.