Re: Index selection bug - Mailing list pgsql-bugs

From Tom Lane
Subject Re: Index selection bug
Date
Msg-id 10680.964621641@sss.pgh.pa.us
Whole thread Raw
In response to Index selection bug  (Andriy I Pilipenko <bamby@marka.net.ua>)
Responses Bug with 'iscachable' attribute (Was: Index selection bug)  (Andriy I Pilipenko <bamby@marka.net.ua>)
List pgsql-bugs
Andriy I Pilipenko <bamby@marka.net.ua> writes:
>     create function func() returns int as 'select 1' language 'sql';

>     set enable_seqscan to 'off';

>     explain select * from t where f = 1;

>       Index Scan using i on t  (cost=0.00..2.01 rows=1 width=4)

>     explain select * from t where f = func();

>       Seq Scan on t  (cost=100000000.00..100000001.34 rows=1 width=4)

Not a bug, because you didn't declare the function 'iscachable'.
For all the system knows, func() is like random() and will return a
different result at every row.  An indexscan can't be used unless it's
safe to fold the function call down to a constant.  See
http://www.postgresql.org/users-lounge/docs/7.0/postgres/sql-createfunction.htm

            regards, tom lane

pgsql-bugs by date:

Previous
From: Andriy I Pilipenko
Date:
Subject: Index selection bug
Next
From: "Michael Richards"
Date:
Subject: Libpq++ memory leak