Received: from sss.sss.pgh.pa.us (sss.pgh.pa.us [209.114.166.2]) by hub.org (8.9.3/8.9.3) with ESMTP id XAA61185 for ; Sun, 24 Oct 1999 23:48:16 -0400 (EDT) (envelope-from tgl@sss.pgh.pa.us) Received: from sss.sss.pgh.pa.us (localhost [127.0.0.1]) by sss.sss.pgh.pa.us (8.9.1/8.9.1) with ESMTP id XAA14208; Sun, 24 Oct 1999 23:47:10 -0400 (EDT) To: Clayton Cottingham cc: "pgsql-sql@postgresql.org" Subject: Re: [SQL] possible bug? In-reply-to: Your message of Mon, 25 Oct 1999 01:22:06 +0000 <3813B0BD.7A65BCEF@smartt.com> Date: Sun, 24 Oct 1999 23:47:10 -0400 Message-ID: <14206.940823230@sss.pgh.pa.us> From: Tom Lane Clayton Cottingham writes: > can anyone confirm this? > create index idx on table using btree(column ); > cluster idx on table; > drop index idx; > seems to drop all permissions and indexes > on the table involved "cluster" deliberately drops all other indexes on the target table, according to the source code. (If this isn't mentioned in the documentation, it should be.) It looks like the implementation method is to build a whole new table, destroy the old, and rename the new into place. That probably explains why the permissions get lost --- I suppose triggers &etc are not carried over either, and heaven help you if you try to cluster a member of an inheritance hierarchy... I have no idea how hard it'd be to make a less destructive version of "cluster". Another item for the TODO list, I guess. regards, tom lane