Re: A few questions:

Christopher Kings-Lynne wrote:
> 1. Did that fix to not allow cluster on partial and non-null indexes get
> backpatched?

Yes, I see it in 7.3.X CVS.

> 2. How can I deliberately cause a deadlock in order to test some code?

Sure:

CREATE TABLE t1(x int);
CREATE TABLE t2(x int);
INSERT INTO t1 VALUES (1);
INSERT INTO t2 VALUES (1);

CLIENT 1:
BEGIN;
UPDATE t1 SET x=2;

CLIENT 2:
BEGIN;
UPDATE t2 SET x=2;
UPDATE t1 SET x=1; -- blocks

CLIENT 1:
UPDATE t2 SET x=2;

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073