Re: VARIANT / ANYTYPE datatype - Mailing list pgsql-hackers

From Jim Nasby
Subject Re: VARIANT / ANYTYPE datatype
Date
Msg-id C99C1182-D167-42F1-BDD8-A48CAFA98F50@nasby.net
Whole thread Raw
In response to Re: VARIANT / ANYTYPE datatype  (Andrew Dunstan <andrew@dunslane.net>)
Responses Re: VARIANT / ANYTYPE datatype  (Andrew Dunstan <andrew@dunslane.net>)
List pgsql-hackers
On May 4, 2011, at 6:24 PM, Andrew Dunstan wrote:
> I'm far from convinced that storing deltas per column rather than per record is a win anyway. I don't have hard
numbersto hand, but my vague recollection is that my tests showed it to be a design that used more space. 

It depends on how many fields you're changing in one go and how wide the table is. It's also a PITA to identify what
fieldsactually changed if you're storing everything. In the case of logging, I'd say that what's really needed is a way
tostore a table record that has an indicator of what fields actually changed (and possibly not storing anything for
fieldsthat didn't change). That table record would need to also deal with changes to the underlying table structure. 
--
Jim C. Nasby, Database Architect                   jim@nasby.net
512.569.9461 (cell)                         http://jim.nasby.net




pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Visibility map and hint bits
Next
From: Andrew Dunstan
Date:
Subject: Re: Patch to improve style of generate_history.pl perl script