The other point about pg_dump failing to correctly restore arrays with nondefault lower bounds is a good one, though. We need to think about how to fix that.
I'll put some thought into it, but note that it is hardly a new issue -- it's been possible to create an array with < 1 lower bound since well before 7.4:
regression=# select version();
version
-----------------------------------------------------------------------
PostgreSQL 7.3.5 on i686-pc-linux-gnu, compiled by GCC gcc (GCC) 3.2.2 20030222 (Red Hat Linux 3.2.2-5)
(1 row)
regression=# create table a(f1 int, f2 int[]);
CREATE TABLE
regression=# insert into a values (1,'{1,2}');
INSERT 565511 1
regression=# update a set f2[0] = 0 where f1 = 1;
UPDATE 1
regression=# select array_dims(f2) from a;
array_dims
------------
[0:2]
(1 row)Joe
---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster
