Re: selecting data from information_schema.columns

From: Jim Nasby <decibel(at)decibel(dot)org>
To: Steve Martin <smartin(at)nec(dot)co(dot)nz>
Cc: pgsql-performance(at)postgresql(dot)org
Subject: Re: selecting data from information_schema.columns
Date: 2006-10-03 01:49:49
Message-ID: F3FACD77-6C3D-4FCA-A346-A3DA238C7CFB@decibel.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

On Oct 2, 2006, at 7:31 PM, Steve Martin wrote:
> Regarding, newsysviews, what is the current state, I have had a
> quick look at the pgFoundry site and the last updates were 9 months
> ago.

Well, the system catalogs don't change terribly often, so it's not
like a lot needs to be done. We'd hoped to get them into core, but
that didn't pan out. Theoretically, we should be making the views
look more like information_schema, but no one's gotten to it yet.

> The most efficient way in the short term I can find to improve
> performance for our application is to create a table from
> information_schema.columns and update it when tables a created or
> deleted, or columns added or removed. E.g.

Well, there's nothing that says you have to use information_schema.
You can always query the catalog tables directly. Even if you don't
want to use newsysviews as-is, the code there should be very helpful
for doing that.

There is no ability to put triggers on DDL, so the best you could do
with your caching table is to just periodically update it.
--
Jim Nasby jim(at)nasby(dot)net
EnterpriseDB http://enterprisedb.com 512.569.9461 (cell)

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Carlo Stonebanks 2006-10-03 02:43:46 Re: Performace Optimization for Dummies
Previous Message Bruno Wolff III 2006-10-03 00:01:47 Re: BUG #2658: Query not using index