Обсуждение: Querying dependencies

Wells Oliver <wells.oliver@gmail.com> writes:
> Sometimes I'll "drop table .." on a replica just to see the list of
> dependencies,.. This is quite dumb. What's the simple query I can run to
> get the same detail without the idiocy?

Not sure about "simple", but here's an example of looking at the
catalog data for this:

regression=# create table mytable(a int primary key, b text unique);
CREATE TABLE
regression=# create table othertable (a int references mytable);
CREATE TABLE
regression=# select pg_describe_object(classid,objid,objsubid), deptype from pg_depend where refclassid =
'pg_class'::regclassand refobjid = 'mytable'::regclass; 
                pg_describe_object                | deptype
--------------------------------------------------+---------
 type mytable                                     | i
 toast table pg_toast.pg_toast_40635              | i
 constraint mytable_pkey on table mytable         | a
 constraint othertable_a_fkey on table othertable | n
 constraint mytable_b_key on table mytable        | a
(5 rows)

See https://www.postgresql.org/docs/current/catalog-pg-depend.html

In the general case you'd need to worry about indirect
dependencies, so you'd need to embed this in a recursive
CTE.  But tables don't usually have indirect dependencies.

            regards, tom lane