Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by postgresql.org (Postfix) with ESMTP id 2F74A474E49 for ; Mon, 25 Feb 2002 23:47:32 -0500 (EST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.11.4/8.11.4) with ESMTP id g1Q4lZS27311; Mon, 25 Feb 2002 23:47:35 -0500 (EST) To: "Josh Berkus" Cc: pgsql-sql@postgresql.org Subject: Re: Timestamp operator error In-reply-to: References: Comments: In-reply-to "Josh Berkus" message dated "Mon, 25 Feb 2002 20:34:36 -0800" Date: Mon, 25 Feb 2002 23:47:35 -0500 Message-ID: <27308.1014698855@sss.pgh.pa.us> From: Tom Lane X-Archive-Number: 200202/378 X-Sequence-Number: 6800 "Josh Berkus" writes: > Any suggestions on an emergency fix for my (production) database? Emergency fix? This operator didn't behave reasonably in 7.1 either (at least not by my definition of reasonable). What exactly would you have us do? >> I suspect this is good ammunition for the argument I've made from >> time to time that we have too many implicit conversions, not too few. > Yes, definitely. Frankly, I'd prefer a large reduction in implicit > conversions; I just got into trouble with the difference between > current_timestamp and current_date that I would have caught much > earlier if Postgres had disallowed the implicit conversion. Yah. Offhand I'd argue that no information-discarding conversion should be implicitly invokable. date->timestamp is fine; timestamp->date should require an explicit cast. I've already proposed that we add a flag to pg_proc to distinguish implicit from explicit conversion operations, and no one complained. But we have not yet begun to argue about exactly which conversions should be allowed implicitly... regards, tom lane