Received: from sss2.sss.pgh.pa.us (sss.pgh.pa.us [209.114.166.2]) by hub.org (8.9.3/8.9.3) with ESMTP id WAA06965 for ; Wed, 19 Apr 2000 22:50:52 -0400 (EDT) (envelope-from tgl@sss.pgh.pa.us) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss2.sss.pgh.pa.us (8.9.3/8.9.3) with ESMTP id WAA00306; Wed, 19 Apr 2000 22:50:25 -0400 (EDT) To: Christopher Sawtell cc: "pgsql-sql@postgreSQL.org" Subject: Re: Counting distinct names In-reply-to: <00042014371500.19753@berty> References: <38FBBD02.4424CB34@nsi.edu> <38FCACFC.55B12EB0@nsi.edu> <22328.956120719@sss.pgh.pa.us> <00042014371500.19753@berty> Comments: In-reply-to Christopher Sawtell message dated "Thu, 20 Apr 2000 14:13:53 +1200" Date: Wed, 19 Apr 2000 22:50:24 -0400 Message-ID: <303.956199024@sss.pgh.pa.us> From: Tom Lane X-Archive-Number: 200004/212 Christopher Sawtell writes: > 1) There are files on the ftp sites which seem to be the "whole thing" > and what I suppose is the "whole thing" cut up into smaller sections. Yes. > Am I correct? Or do I need to d/l the smaller files as well? > A README_for_Download might be a good idea. Yo, Marc ... where's the documentation about the new tarfile sections? > 2) I couldn't find a --database-dir=[/var/lib/pgsql/data] option in > the output from ./configure --help. How can one configure pg to place > it's database in a defined place in the file system? btw the default > of /var/lib/pgsql/data seems pretty sensible. The default location is PREFIX/data; I don't think there is a configure option to change that without setting --prefix. However, you can set it at runtime with the -D options to initdb and postmaster, so adding yet another configure switch might be gilding the lily. > 3) Some of the scripts and binary files for the subsiduary tools have > to be installed by hand. If you need to know what, I'd be happy to > re-install pg and take notes. News to me, other than that the perl interface module usually has to be installed separately because of permission issues. What did you see missing? > 5) Is the idea of putting the data on a raw partition on the horizon? Not really. There are portability issues with that; furthermore, it's not obvious that it'd be better. The system's been developed and tuned to sit atop a Unix kernel's disk cache. Writing our own disk management code would be a large investment of work and a large loss of portability for very questionable benefits. It might happen someday, but I think we have many higher-priority tasks ahead... regards, tom lane