Optimize Oracle introspection by using USER_SEQUENCES instead of USER_CATALOG

[Mariusz Felisiak]

Little optimization. Using USER_SEQUENCES instead of USER_CATALOG in Oracle query which verify that sequence exists or not is more optimal.

[Simon Charette]

Please don't mark your own patches as RFC, see ​our ticket triaging doc's.

[Simon Charette]

​PR

[Tim Graham]

Could you explain why it's more optimal so we have that rationale recorded?

[Mariusz Felisiak]

Replying to Tim Graham:

Could you explain why it's more optimal so we have that rationale recorded?

Query with USER_CATALOG has a cost equal 3 and it has 2 conditions in WHERE clause.
Query with USER_SEQUENCES has a cost equal 2 and it has 1 condition in WHERE clause.

USER_CATALOG is quite complicate, it depends on the _CURRENT_EDITION_OBJ and it contains all of the user objects from which we are choosing sequences:

CREATE OR REPLACE FORCE VIEW "SYS"."_CURRENT_EDITION_OBJ" ("OBJ#", "DATAOBJ#", "DEFINING_OWNER#", "NAME", "NAMESPACE", "SUBNAME", "TYPE#", "CTIME", "MTIME", "STIME", "STATUS", "REMOTEOWNER", "LINKNAME", "FLAGS", "OID$", "SPARE1", "SPARE2", "SPARE3", "SPARE4", "SPARE5", "SPARE6", "OWNER#", "DEFINING_EDITION") AS 
  select o."OBJ#",o."DATAOBJ#",o."OWNER#",o."NAME",o."NAMESPACE",o."SUBNAME",o."TYPE#",o."CTIME",o."MTIME",o."STIME",o."STATUS",o."REMOTEOWNER",o."LINKNAME",o."FLAGS",o."OID$",o."SPARE1",o."SPARE2",o."SPARE3",o."SPARE4",o."SPARE5",o."SPARE6",
       o.spare3,
       case when (o.type# not in (4,5,7,8,9,10,11,12,13,14,22,87) or
                  bitand(u.spare1, 16) = 0) then
         null
       when (u.type# = 2) then
        (select eo.name from obj$ eo where eo.obj# = u.spare2)
       else
        'ORA$BASE'
       end
from obj$ o, user$ u
where o.owner# = u.user#
  and (   /* non-versionable object */
          (   o.type# not in (4,5,7,8,9,10,11,12,13,14,22,87,88)
           or bitand(u.spare1, 16) = 0)
          /* versionable object visible in current edition */
       or (    o.type# in (4,5,7,8,9,10,11,12,13,14,22,87)
           and (   (u.type# <> 2 and
                    sys_context('userenv', 'current_edition_name') = 'ORA$BASE')
                or (u.type# = 2 and
                    u.spare2 = sys_context('userenv', 'current_edition_id'))
                or exists (select 1 from obj$ o2, user$ u2
                           where o2.type# = 88
                             and o2.dataobj# = o.obj#
                             and o2.owner# = u2.user#
                             and u2.type#  = 2
                             and u2.spare2 =
                                  sys_context('userenv', 'current_edition_id'))
               )
          )
      );

USER_SEQUENCES is much more readable and it contains only sequences:

CREATE OR REPLACE FORCE VIEW "SYS"."USER_SEQUENCES" ("SEQUENCE_NAME", "MIN_VALUE", "MAX_VALUE", "INCREMENT_BY", "CYCLE_FLAG", "ORDER_FLAG", "CACHE_SIZE", "LAST_NUMBER") AS 
  select o.name,
      s.minvalue, s.maxvalue, s.increment$,
      decode (s.cycle#, 0, 'N', 1, 'Y'),
      decode (s.order$, 0, 'N', 1, 'Y'),
      s.cache, s.highwater
from sys.seq$ s, sys.obj$ o
where o.owner# = userenv('SCHEMAID')
  and o.obj# = s.obj#;

To sum up, in the first approach we choose all objects (including sequences), and in the second step we limit results to sequences with the given name. In the second approach we choose sequences and next limit results to the given name.

[Shai Berger]

I verified that USER_SEQUENCES is available on Oracle 11, I don't really care if it is as optimized there.

I left one comment on the PR, otherwise I think we're good to go.

[Tim Graham <timograham@…>]

In 794f866c:

Fixed #27323 -- Optimized Oracle introspection by using USER_SEQUENCES instead of USER_CATALOG.