Hi, I've already implemented preserving PG_PROC oids during pg_upgrade in a way like relfilenodes, etc, actually, it is quite simple, and on the first look there are no any problems.
About using surrogate key - this feature is more for data generated by the DBMS itself, i.e. data processed by some extension and saved and re-processed automatically or by user's request, but without bothering user with these internal keys. The main question - maybe, are there pitfalls of which I am not aware of? Thanks for your replies! -- Regards, Nikita Malakhov Postgres Professional The Russian Postgres Company https://postgrespro.ru/
