From: | Joe Conway <mail(at)joeconway(dot)com> |
---|---|
To: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: select t.name from tbl t (where "name" is not a column name) |
Date: | 2010-02-24 02:18:59 |
Message-ID: | 4B848C93.1080900@joeconway.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On 02/23/2010 05:07 PM, raf wrote:
> i've just noticed the following behaviour and was wondering
> if there's any documentation to explain what it's for.
>
> create table tbl(id serial primary key, a text, b text, c text);
> insert into tbl(a, b, c) values ('abc', 'def', 'ghi');
> insert into tbl(a, b, c) values ('jkl', 'mno', 'pqr');
> insert into tbl(a, b, c) values ('stu', 'vwx', 'yza');
> select t.name from tbl t;
I forget exactly where this is documented (and could not find it with a
quick look), but calling t.name is the same as name(t) if a column
reference is not found, and name is a function, which it is.
So t.name is essentially casting the whole row as a name datatype and
outputting the result. Try it with text:
test=# \d foo
Table "public.foo"
Column | Type | Modifiers
--------+---------+-----------
f | integer |
test=# select foo.text from foo;
text
------
(-1)
(1 row)
test=# drop TABLE foo;
DROP TABLE
test=# create table foo(f int, text text);
CREATE TABLE
test=# insert into foo values(-1,'abc');
INSERT 0 1
test=# select foo.text from foo;
text
------
abc
(1 row)
test=# select foo.name from foo;
name
----------
(-1,abc)
(1 row)
HTH,
Joe
From | Date | Subject | |
---|---|---|---|
Next Message | Baron Schwartz | 2010-02-24 04:08:22 | Re: Sorting performance vs. MySQL |
Previous Message | Net Tree Inc. | 2010-02-24 01:57:14 | Re: how do I do dump and restore without bugging with constraint? |