From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Scott Marlowe <smarlowe(at)g2switchworks(dot)com> |
Cc: | Alex Turner <armtuk(at)gmail(dot)com>, Wiebe de Jong <wiebedj(at)shaw(dot)ca>, pgsql-general(at)postgresql(dot)org |
Subject: | Re: Determining when a row was inserted |
Date: | 2005-06-03 19:26:30 |
Message-ID: | 12132.1117826790@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Scott Marlowe <smarlowe(at)g2switchworks(dot)com> writes:
> On Fri, 2005-06-03 at 13:40, Alex Turner wrote:
>> One might even suggest that this should really be a default for all
>> tables everywhere, because at some time or another, someone wants to
>> know when something got put in the database...
> That kind of designing is what leads to bloated, overweight programs...
Agreed --- it is much more important to be sure that we have the
features needed to let people add these sorts of behaviors for
themselves (in this case, triggers).
As it happens, the original Berkeley-era Postgres did indeed add
creation and deletion timestamps to every row, as part of their "time
travel" feature. That got ripped out very soon after the code left
Berkeley, because the overhead was just unacceptable ... and our
threshold for unacceptable performance was a whole lot higher then
than it is today ...
It's worth noting in connection with this Joe Hellerstein's description
of Berkeley-era Postgres:
http://archives.postgresql.org/pgsql-hackers/2002-06/msg00085.php
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Brad Nicholson | 2005-06-03 19:31:35 | Re: postgresql books |
Previous Message | Scott Marlowe | 2005-06-03 19:19:23 | Re: Determining when a row was inserted |