Re: index suggestion for 7.4
Bruno Wolff III <bruno@wolff.to>
From: Bruno Wolff III <bruno@wolff.to>
To: Stephan Szabo <sszabo@megazone23.bigpanda.com>
Cc: pgsql-hackers@postgresql.org
Date: 2003-05-30T18:02:13Z
Lists: pgsql-hackers
On Fri, May 30, 2003 at 10:42:24 -0700, Stephan Szabo <sszabo@megazone23.bigpanda.com> wrote: > On Fri, 30 May 2003, Bruno Wolff III wrote: > > > Now that expressions can be used in indexes in 7.4 you can have multicolumn > > indexes that are ordered in different directions. However the planner > > doesn't seem to understand that order by -col asc is the same as order by > > col desc (for at least the normal -) so you have to be careful how you > > write queries when doing this. > > I think it'd be better to make it easier to make indexes where some > columns are reversed. I'm not sure that making a reverse opclass for > btree (one that goes >, >=, =, <=, < I guess) is a complete solution > even for btree but if it is, we could provide them. I think this would > also have the advantage of not requiring wacky queries to use the index > for multicolumn lookups as well. I was hoping the new stuff Tom added would make doing this easier. The issue has come up before and at least at that time it didn't get changed so I expected it wasn't easy to do. I thought maybe there was information for the - operator that would allow you to know that you could use an index on -col to go in the reverse direction safely. The new stuff still is easier to use then creating a new opclass which was the old solution.