Re: feature proposal ...

From: Rod Taylor <pg(at)rbt(dot)ca>
To: tshipley(at)deru(dot)com
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: feature proposal ...
Date: 2005-09-21 22:32:23
Message-ID: 1127341943.94955.40.camel@home
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Wed, 2005-09-21 at 15:25 -0700, Trent Shipley wrote:
> On Wednesday 2005-09-21 07:01, Hans-Jürgen Schönig wrote:
> > Rod Taylor wrote:
> > >>the problem is: COPY can write data returned by a SELECT statement to a
> > >>file. our idea is to implement precisely that.
> > >>
> > >>example:
> > >>
> > >>COPY TO file_name USING some_select_statement;
> > >
> > > I have run into plenty of cases where I wanted to dump part of a
> > > structure and this could be used for that, but I've always found that
> > > temporary tables were sufficient and equally SQL scriptable
> > >
> > > CREATE TEMP TABLE tab AS SELECT ...; COPY tab TO file_name;
> >
> > Hi Rod,
> >
> > TEMP TABLE are not suitable for my case. Using a temp table would
> > essentially mean that we had to store the data 3 times: Original data,
> > temp table + dump. Temp tables are only fine for small amounts of data
> > but we are talking about too much data here (my smallest export will
> > contain 15.000.000 records).
>
> Wouldn't you also need a CREATE TEMP TABLE privilege but the
> COPY TO file USING select_statement
> would only need select. (In other words using a temp table would not seem to
> be as secure nor as general as the requested feature.)

Writing a file on the server requires significant privilege, including
access to the server itself so you can retrieve the results.
--

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Andrew Dunstan 2005-09-21 23:55:38 Re: feature proposal ...
Previous Message Trent Shipley 2005-09-21 22:25:47 Re: feature proposal ...