I have:
psql (PostgreSQL) 7.3.2
I do a modification of 'access/index/indexam.c' where I comment:
#ifdef NOT_USED
        if (scan->keys_are_unique && scan->got_tuple)
        {
                if (ScanDirectionIsForward(direction))
                {
                        if (scan->unique_tuple_pos <= 0)
                                scan->unique_tuple_pos++;
                }
                else if (ScanDirectionIsBackward(direction))
                {
                        if (scan->unique_tuple_pos >= 0)
                                scan->unique_tuple_pos--;
                }
                if (scan->unique_tuple_pos == 0)
                        return heapTuple;
                else
                        return NULL;
        }
#endif

I do not remember the references of the bug.
But the solution was planned for 7.4.

I do:
psql=# \di
[skip]
 public | url_next_index_time      | index | postgresql | url
 [skip]
(11 rows)
I have an index on next_index_time field on table url.

psql=# explain select min(next_index_time) from url \g
                            QUERY PLAN
-------------------------------------------------------------------
 Aggregate  (cost=85157.70..85157.70 rows=1 width=4)
   ->  Seq Scan on url  (cost=0.00..80975.56 rows=1672856 width=4)
(2 rows)
Silly SeqScan of all the table.

psql=# explain SELECT next_index_time FROM url ORDER BY next_index_time LIMIT 1 \g
QUERY PLAN
----------------------------------------------------------------------- -------------------------
Limit (cost=0.00..0.20 rows=1 width=4)
-> Index Scan using url_next_index_time on url (cost=0.00..340431.47 rows=1672856 width=4)
(2 rows)
I ask for the same thing.
That's better !

Why the planner does that ?

Jean-Gérard Pailloncy
Paris, France


---------------------------(end of broadcast)--------------------------- TIP 3: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly

Reply via email to