Received: from localhost (postgresql.org [64.49.215.8]) by postgresql.org (Postfix) with ESMTP id D52D4476CD5 for ; Mon, 13 Jan 2003 16:33:16 -0500 (EST) Received: from o2.hostbaby.com (o2.hostbaby.com [208.187.29.121]) by postgresql.org (Postfix) with SMTP id 3165E476B4D for ; Mon, 13 Jan 2003 16:33:16 -0500 (EST) Received: (qmail 32631 invoked from network); 13 Jan 2003 21:33:19 -0000 Received: from unknown (HELO l-i-e.com) (127.0.0.1) by localhost with SMTP; 13 Jan 2003 21:33:19 -0000 Received: from 216.80.95.13 (Hostbaby Webmail authenticated user typea@l-i-e.com) by www.l-i-e.com with HTTP; Mon, 13 Jan 2003 13:33:19 -0800 (PST) Message-ID: <49456.216.80.95.13.1042493599.squirrel@www.l-i-e.com> Date: Mon, 13 Jan 2003 13:33:19 -0800 (PST) Subject: Cursor rowcount From: To: X-Priority: 3 Importance: Normal X-Mailer: Hostbaby Webmail MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by AMaViS new-20020517 X-Archive-Number: 200301/77 X-Sequence-Number: 724 Short Version: I've read the idocs and Notes and Googled a fair amount, honest. :-) What's the most efficient way of determining the number of rows in a cursor's result set if you really *DO* need that? (Or, rather, if your client specifically asked for features that require that.) Long Version: I'm not finding any way in the docs of asking a cursor how many rows total are in the result set, even if I do "move 1000000 in foo", knowing a priori that 1000000 is far more than could be returned. Oracle docs seem to have a SQL.%ROWCOUNT which gives the answer, provided one has moved beyond the last row... If I'm reading the Oracle docs right... Anyway. I could find nothing similar in PostgreSQL, even though it seems reasonable, even for a Portal, provided one is willing to do the "move X" for X sufficiently high -- And, in fact, psql outputs the precise number of rows when I do that in the psql monitor, so at some level PostgreSQL "knows" the answer I want, but I can't get that "MOVE XX" output into PHP, as far as I can tell. (Can I?) I suppose I could, in theory, use PHP to fire up psql, but that's not exactly going to be efficient, much less pleasant. :-) Using PHP, if it matters. I guess it does since maybe other APIs have some way to access that number I want -- psql sure seems to print it out when one goes over the edge. Given that the count(*) queries take just as long as the actual data-retrieval queries, and that some of my queries take too long as it is (like, a minute for a 4-term full text search)... I've written and am about to benchmark a binary search using a bunch of "move X" "fetch 1" "move backward 1" "move backward X" and then using Ye Olde Low/High guessing game algorithm to find the number of rows, but I'm hoping for something better from the optimization experts. Sorry this got a bit long, but I wanted to be clear about where I've been and gone, rather than leave you guessing. :-) Hope I didn't miss some obvious solution/documentation "out there"...