Re: Measure Theoretic Data Types in Postgresql

From: Josh Berkus <josh(at)agliodbs(dot)com>
To: pgsql-hackers(at)postgresql(dot)org
Subject: Re: Measure Theoretic Data Types in Postgresql
Date: 2012-10-12 17:06:41
Message-ID: 50784E21.2030104@agliodbs.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 10/12/12 12:48 AM, Heikki Linnakangas wrote:
>
> So, I think the current range types already cover that use case pretty
> well. I can't imagine how the proposed measure theoretic concepts would
> make that simpler. Can you give some more complicated problem, perhaps,
> that the proposed measure theoretic concepts would make simpler than the
> current tools?

Well, the nice thing about EXTENSIONs is that, if he builds it, people
can just try it out and see if it's useful. I suspect that the use
cases are rarified enough that this would always be an EXTENSION and not
core.

One thing I could use, for example, would be a time-keyed array, in the
form:

metrics_series ( '2012-10-17 45:22:10',10,
'1 second',15.0,15.1,16.2,NULL,15.8, 14.9, 15.1,14.2, 13.9, NULL )

WHich would allow me to do:

SELECT metric WHERE ts = '2012-10-17 45:22:14'

Without storing a timestamp with each element.

This is not a set of functionality I would expect to be generally useful
and belong in Core. But for a certain set of analytics applications, it
would be indispensible.

I expect that theoretic data types are the same way.

--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Pavel Stehule 2012-10-12 17:27:25 [PATCH] assign result of query to psql variable
Previous Message Tom Lane 2012-10-12 17:05:44 Re: [PATCH] explain tup_fetched/returned in monitoring-stats