Japin Li <japinli@hotmail.com> writes:
> I found the misc_sanity has a SQL to check system catalogs that
> do not have primary keys, however, in current exceptions it says
> pg_depend, pg_shdepend don't have a unique key.
As indeed they do not:
regression=# \d pg_depend
Table "pg_catalog.pg_depend"
Column | Type | Collation | Nullable | Default
-------------+---------+-----------+----------+---------
classid | oid | | not null |
objid | oid | | not null |
objsubid | integer | | not null |
refclassid | oid | | not null |
refobjid | oid | | not null |
refobjsubid | integer | | not null |
deptype | "char" | | not null |
Indexes:
"pg_depend_depender_index" btree (classid, objid, objsubid)
"pg_depend_reference_index" btree (refclassid, refobjid, refobjsubid)
regression=# \d pg_shdepend
Table "pg_catalog.pg_shdepend"
Column | Type | Collation | Nullable | Default
------------+---------+-----------+----------+---------
dbid | oid | | not null |
classid | oid | | not null |
objid | oid | | not null |
objsubid | integer | | not null |
refclassid | oid | | not null |
refobjid | oid | | not null |
deptype | "char" | | not null |
Indexes:
"pg_shdepend_depender_index" btree (dbid, classid, objid, objsubid), tablespace "pg_global"
"pg_shdepend_reference_index" btree (refclassid, refobjid), tablespace "pg_global"
Tablespace: "pg_global"
Your proposed wording seems to give strictly less information,
so I do not see why it's an improvement.
regards, tom lane