Postgres

Every chart and table on this page reads from a live Postgres database. data.py attaches it as the catalog pg, and each query below runs in your browser against that connection, over the real pg.public.users and pg.public.orders, with no export step and no local copy.

At a glance

Breakdowns

orders_by_status and users_by_country are SQL views defined in data.py with @pg.view(). They evaluate in your browser against the attached tables; nothing is copied into a local database.

Editable orders

data.py exposes public.orders as "readwrite". Edit a cell or add a row and the change writes back to Postgres over the same connection: the UPDATE or INSERT runs against the real table, not a local copy.

Read-only users

public.users is exposed as "read", so this grid browses but never writes.

Row-level security

Signed in, you see only your own orders, and that's enforced on the server, never in the browser. login records who you are on the session, and my_orders() takes no user id at all. It scopes every read to the identity login already set, so no SQL you type in the browser can widen your view to another user's orders.

Log in below. The toggle calls login, and the same my_orders() returns only that user's rows, nobody's while you're logged out.