Hello
I am not sure it is that simple. Probably you need to create operator classes to be used for indexing.
http://www.postgresql.org/docs/9.4/static/xtypes.html
You are probably better off using the basic data type in your table and using a composite index.
Bye
Charles
From: pgsql-general-owner@xxxxxxxxxxxxxx [mailto:pgsql-general-owner@xxxxxxxxxxxxxx] On Behalf Of John R Pierce
Sent: Montag, 29. Juni 2015 07:51
To: pgsql-general@xxxxxxxxxxxxxx
Subject: Re: create index on a field of udt
On 6/28/2015 10:31 PM, Shujie Shang wrote:
Oh, I didn't explain my question well, actually I want to create an index on an udt in a table.
e.g.
create type info as (id int, name text);
creat table test (i info);
I want to run:
create index myindex on test (i.id)
create table test of info primary key(id);
or, if you want to use your type plus other stuff in the table, I believe its something like...
create table test (i info, stuff...) primary key (i.id)
or
create index test(i.id);
watch out for ambiguity if the type names match the table or field name. see http://www.postgresql.org/docs/current/static/rowtypes.html#AEN7836--john r pierce, recycling bits in santa cruz
Hi,
I find a way to create index, I create a function returns the 'id' field of udt info, then I create index based on this function.
e.g
create type info as (id int, name text);
creat table test (id int, i info);
create or replace function getID(i info) returns int as
$$ select $1.id $$
language sql;
create index infoindex on test (getID(i));
I want to use this index, but after I insert lots of data to the table 'test' and run 'select * from test where i.id=5', it still use 'seqscan', not 'index scan'. How can I verify the index is build correctly?
e.g.
insert into test values (generate_series(1, 3000000), (1, 'hi')::info);
explain select * from test where i.id=1;
the result is : seqscan
On Mon, Jun 29, 2015 at 1:57 PM, Charles Clavadetscher <clavadetscher@xxxxxxxxxxxx> wrote:
