a btree strategy hosed in 6.4.BETA1 and in current snapshot
От | Gene Selkov, Jr. |
---|---|
Тема | a btree strategy hosed in 6.4.BETA1 and in current snapshot |
Дата | |
Msg-id | 199810102104.QAA20861@antares.mcs.anl.gov обсуждение исходный текст |
Список | pgsql-hackers |
Hi hackers, I have trouble understanding the message I got in response to CREATE INDEX: CREATE INDEX ecix ON ec USING btree ( ec ec_code_ops ); ERROR: RelationInvokeStrategy: cannot evaluate strategy 5 The strategy in question is defined as: INSERT INTO pg_amop (amopid, amopclaid, amopopr, amopstrategy, amopselect, amopnpages) SELECT am.oid, opcl.oid, c.opoid, 5, 'btreesel'::regproc, 'btreenpage'::regproc FROM pg_am am, pg_opclass opcl, ec_code_ops_tmp c WHERE amname = 'btree' and opcname = 'ec_code_ops' and c.oprname = '>'; where ec_code_ops_tmp is the result of: SELECT o.oid AS opoid, o.oprname INTO TABLE ec_code_ops_tmp FROM pg_operator o, pg_type t WHERE o.oprleft = t.oid and o.oprright = t.oid and t.typname = 'ec_code'; But I can evaluate this function as an operator without problems: emp=> select * from test_ec where ec > '5'; ec --------- 6.2.-.- 5.4.1.9 5.4.3.9 5.2.1.114 (4 rows) This did not happen to me in older versions, including 6.3.2-release. I am wondering whether something changed in btree after6.3.2, and how do I accomodate my extensions if it did. Thank you, Gene (I would appreciate a direct reply, as I am not a member of the hackers list)
В списке pgsql-hackers по дате отправления: