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 PAA28595; Fri, 10 Sep 1999 15:07:52 -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 PAA10808; Fri, 10 Sep 1999 15:07:18 -0400 (EDT) To: Stuart Rison cc: pgsql-general , pgsql-sql Subject: Re: [GENERAL][SQL] Getting multiple field unique index to distinguish NULLs. In-reply-to: Your message of Fri, 10 Sep 1999 16:37:25 +0100 (BST) Date: Fri, 10 Sep 1999 15:07:18 -0400 Message-ID: <10806.936990438@sss.pgh.pa.us> From: Tom Lane Stuart Rison writes: > Now this would work except that there's a problem with pl/pgSQL such that > when you pass several values to a plpgsql function and one of them is > NULL, then all values passed to the function become NULL. Yes. That's not plpgsql's fault; there's a basic design limitation in the function-call interface (only one isnull flag gets passed). There has been some talk on the pghackers list of fixing this, but I do not know if it will happen any time soon or not. Fixing that interface would mean revising *all* the code that implements any SQL function or operator, so it'd be pretty tedious --- and it'd break any C-coded user functions, too. It'll probably get done eventually, maybe even for 6.6, but don't hold your breath... regards, tom lane