Consider the following:

create table foo ( id serial primary key, path text);
alter table foo add ref foo;

Table "public.foo"
Column | Type | Modifiers
--------+---------+-----------------------------------------------------
id | integer | not null default nextval('public.foo_id_seq'::text)
path | text |
ref | foo |
Indexes:
"foo_pkey" PRIMARY KEY, btree (id)


insert into foo (path) VALUES ('/');
insert into foo (path,ref) VALUES ('/foo',ROW(1,'/',NULL));
insert into foo (path,ref) VALUES
('/foo/bar',ROW(2,'/foo',ROW(1,'/',NULL)));

select * from foo;
id | path | ref
----+----------+-------------------
1 | / |
2 | /foo | (1,/,)
3 | /foo/bar | (2,/foo,"(1,/,)")
(3 rows)

However, it is not possible to create such a type using create type, as
there is no alter type, even tho alter table does the same thing.

Is this a bug or a random feature?

It is also not possible to dump and restore this using pg_dump.

.... John

---------------------------(end of broadcast)---------------------------
TIP 7: don't forget to increase your free space map settings

"John Hansen" <john@geeknet.com.au> writes:
> create table foo ( id serial primary key, path text);
> alter table foo add ref foo;

> Is this a bug or a random feature?

I'd say it's a random misfeature ;-). I certainly don't care to promise
that it's supported ... but on the other hand, preventing you from doing
it seems nontrivial (consider indirect recursions). I'd put this in the
category of "doctor, it hurts when I do that".

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 3: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to majordomo@postgresql.org so that your
message can get through to the mailing list cleanly