Date: Fri, 5 Nov 2021 10:17:12 -0400
From: Bruce Momjian <bruce@momjian.us>
To: Laurenz Albe <laurenz.albe@cybertec.at>
Cc: Grega Jesih <Grega.Jesih@actual-it.si>,
	"David G. Johnston" <david.g.johnston@gmail.com>,
	Tom Lane <tgl@sss.pgh.pa.us>,
	Pg Docs <pgsql-docs@lists.postgresql.org>
Subject: Re: text fields and performance for ETL
Message-ID: <20211105141712.GA19812@momjian.us>
References: <163594615980.678.1904941309603792278@wrigleys.postgresql.org>
 <CAKFQuwa_02fEtL8zJMt7MR=xvrp-De1pQtHFtg7fSHhSm1ajnA@mail.gmail.com>
 <686716cb5c4242c28e5967d9b6fcfc75@actual-it.si>
 <137264.1635961048@sss.pgh.pa.us>
 <c57e7a702932465ca9962d4f5d332c3a@actual-it.si>
 <CAKFQuwYFfESp=xUpktYBbW8TxO8k=+uWN8F2G5Ehis2cpYuDnw@mail.gmail.com>
 <70cf9fe110c246afb4eab054caab0c04@actual-it.si>
 <CAKFQuwbX37AwNcKHMeaOq+LS_nhw_6sHwaUK1+UsQ8Nuo7bcfg@mail.gmail.com>
 <516dc58b80fa418091f5262879019213@actual-it.si>
 <b288134ae2290d5abc2a2d516a8d58892a656f30.camel@cybertec.at>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <b288134ae2290d5abc2a2d516a8d58892a656f30.camel@cybertec.at>
User-Agent: Mutt/1.10.1 (2018-07-13)
Archived-At: 
 <https://www.postgresql.org/message-id/20211105141712.GA19812%40momjian.us>
Precedence: bulk

On Fri, Nov  5, 2021 at 03:15:35PM +0100, Laurenz Albe wrote:
> On Fri, 2021-11-05 at 09:51 +0000, Grega Jesih wrote:
> > Suggested current text addendum:
> > 
> > But, if you consider doing ETL from Postgres database to some outer target
> > environment and you seek performance in such interfaces, follow the logic
> > of limited size (varchar or char) data types in your database model.
> > Because if you make a dataflow of known size types, interface code can take
> > a big block of data while for text fields you need to check each record.
> 
> I am opposed to that.
> 
> It is not our business to discuss the limitations of a certain third-party software product.
> If that were something wide-spread, perhaps.  But I myself have never seen a problem
> with "text", as long as the actual size of the data is moderate.

Agreed.

> > Optional additional remark:
> > 
> > Another good aspect of known data sizes is easier understanding of field
> > content and implicit data (length) control.
> 
> Something like that makes more sense to me.
> 
> Perhaps, right before the tip you quoted, something like that:
> 
>   If your use case requires a length limit on character data, or compliance
>   with the SQL standard is important, use "character varying".
>   Otherwise, you are usually better off with "text".

I can support that if others think it is valuable.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  If only the physical world exists, free will is an illusion.