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 JAA16338 for ; Tue, 28 Sep 1999 09:53:37 -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 JAA00404; Tue, 28 Sep 1999 09:51:49 -0400 (EDT) To: Bruce Momjian cc: Howie , pgsql-sql@hub.org Subject: Re: [SQL] table aliasing problem with 6.5... In-reply-to: Your message of Tue, 28 Sep 1999 00:48:35 -0400 (EDT) <199909280448.AAA01722@candle.pha.pa.us> Date: Tue, 28 Sep 1999 09:51:48 -0400 Message-ID: <402.938526708@sss.pgh.pa.us> From: Tom Lane Bruce Momjian writes: >> In a situation where you've got subselects, it may not be immediately >> obvious which FROM list the entry got added to. I can't think of any >> simple way of identifying that, however :-( > Not sure how to handle that either. I could print the subquery level > number, or I could say "in subquery" or even go fancy and do "in > sub-sub-query" depending on the number of levels down. That seems like a good idea; it won't help tell the difference between two subqueries at the same level, but in a lot of practical cases it would tell you what you needed to know, and it won't confuse a novice. I like the "... in subquery", "... in sub-subquery", etc wording. BTW, Jan was complaining that a number of the regress tests now "fail" because they provoke this message. Should we just update the expected outputs, or should we change the tests not to use the feature? regards, tom lane