hstore does look like it could work out better then a mother load of columns like we have now so long as the index support is good and there are no limits on value length.
It seems that the limit was lifted in 9.0? Is that true?
I was not aware of hstore when we started using postgres, that’s for the info!
And yes we do data analysis that tortures SQL, but SQL allows us to do so many things quickly and less painfully. Better to torture the machines then torture ourselves….
- Mark
From: pgsql-general-owner@postgresql.org [mailto:pgsql-general-owner@postgresql.org] On Behalf Of Dmitriy Igrishin
Sent: Friday, November 12, 2010 4:50 AM
To: John R Pierce
Cc: pgsql-general@postgresql.org
Subject: Re: [GENERAL] More then 1600 columns?
Sounds like semi-structured data handling. For this tasks hstore would be ideal, IMO.
2010/11/12 John R Pierce <pierce@hogranch.com>
I looked up the OP's domain. They develop medical research data analysis software.
That sort of software and the kinds of data analysis they do tortures SQL databases. These 1600+ element rows are likely very sparse.
--
--
// Dmitriy.
