Re: redundants indexes can be created

hubert depesz lubaczewski wrote:
> On Fri, Jul 04, 2008 at 11:54:37AM +0200, Cyril SCETBON wrote:
>
>> Why postgres does not disallow creating redundants indexes ? Is it the
>> same behaviour in postgresql 8.3 ?
>>
>
> Why should it? Redundant indexes are not "bugs". And can be very useful
> sometimes (thing concurrent reindexing).
>
in this case your right, but lot of people are confused with primary key
and unique key. So they create a unique key on the same column that
constitute the primary key.  For example, Oracle inhib it :

SQL> create table toto(id int primary key);

Table created.

SQL> create unique index idx_toto_id on toto(id);
create unique index idx_toto_id on toto(id)
                                        *
ERROR at line 1:
ORA-01408: such column list already indexed

concurrent reindexing is the matter of postgresql, it should create it
transparently when needed. And if I take into account the concurrent
reindexing, why permitting more than 2 index on the same column ?

--
Cyril SCETBON