Re: full featured alter table? - Mailing list pgsql-general

From Mike Mascari
Subject Re: full featured alter table?
Date
Msg-id 3EEC5D53.9070905@mascari.com
Whole thread Raw
In response to Re: full featured alter table?  (Ernest E Vogelsinger <ernest@vogelsinger.at>)
Responses Re: full featured alter table?  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
Ernest E Vogelsinger wrote:

> At 12:33 15.06.2003, Nigel J. Andrews said:
> --------------------[snip]--------------------
>
>>Well, on the basis that column ordering is presentation issue it does
>>belong in
>>the frontend. However, as Tom pointed out this discussion was had a few
>>months. I can't remember the outcome but apparently a suggestion was to have a
>>column in pg_attribute giving the presentational order.
>
> --------------------[snip]--------------------
>
> Holdit.
>
> This whole thread eveolves to something asking a car designer to design a
> car in a way a particular color would come best...
>
> I believe column ordering should be decided by the database itself, to
> maximize output/throughput or otherwise beneficially influent performance
> and/or disk usage. If you want a specific column sequence, put it into the
> SELECT statement. '*' just means "gimme all of them", not in a particular
> order.

The proposal does beg the question:

Why would a default tuple-attribute order be stored in the database
but not relation-tuple order?

Mike Mascari
mascarm@mascari.com



pgsql-general by date:

Previous
From: Ernest E Vogelsinger
Date:
Subject: Re: full featured alter table?
Next
From: "Robert C. Paulsen Jr."
Date:
Subject: Re: crypt vs password in pg_hba.conf