I’m not sure I understand — can you elaborate on “what’s wrong with an information schema and an order by”? I don’t see how that’s connected to the problems I described with pg_dump or the solution offered by pgmigrate, but I’d be curious to learn another way of solving be problem.

Just the problem with ordering of state in a database not matching the other database after you ran migrations. Information schema being the authoritative schema for all the information about the meta state in most database systems (or at least the cross platform one) means you can write a query against these views and (if I am not mistaken) re-assemble state that's identical by ordering the application of the objects in the order they were created originally.

Is everything in pgdump in the information schema?