|From:||Vlad Arkhipov <arhipov(at)dc(dot)baikal(dot)ru>|
|Cc:||pgsql-hackers(at)postgresql(dot)org, Miroslav Šimulčík <simulcik(dot)miro(at)gmail(dot)com>|
|Subject:||Re: Temporal features in PostgreSQL|
|Views:||Raw Message | Whole Thread | Download mbox | Resend email|
On 02/15/2013 10:46 PM, Cédric Villemain wrote:
> I'm also interested in this topic.
> > > I'm also interested in this topic and work on system-time temporal
> > > extension. Here I wrote down design of my solution few months ago
> > > https://wiki.postgresql.org/wiki/SQL2011Temporal. The idea is
> > > basically the same as in your solution with some minor differences.
> I've added a requirement in the system here: the table to be versioned
> must have a PK (I dislike _entry_id usage but this sounds good othwise).
> I then define a "EXCLUDE WITH GIST (pk with =, sys_period with &&)", thus
> getting expected UNIQUEness also in the history.
I use similar constraints for application-time period tables but not for
system versioned. Because they are automatically controlled by a
trigger, there should be no need for additional integrity checks. If you
want to speed up queries against historical data, you can create GIST
index or an exclusion constraint.
> Vlad, is your source code in a public versionning system (github,
> bucket, etc) ?
> It will ease the process to participate to your extension...
Yes, I uploaded it on github
The extension is also available on PGXN
> Cédric Villemain +33 (0)6 20 30 22 52
> PostgreSQL: Support 24x7 - Développement, Expertise et Formation
|Next Message||Amit Kapila||2013-02-18 04:07:12||Re: [PATCH] Add PQconninfoParseParams and PQconninfodefaultsMerge to libpq|
|Previous Message||Tomas Vondra||2013-02-17 23:49:17||Re: PATCH: Split stats file per database WAS: autovacuum stress-testing our system|