Re: SELECT creates millions of temp files in a single directory

On 4/23/22 14:58, Peter wrote:
> On Sat, Apr 23, 2022 at 02:11:00PM -0700, Adrian Klaver wrote:
> ! On 4/23/22 12:50, Peter wrote:
> !
> !
> ! > People seem to have been brainwashed by Web-Services and OLTP,
> ! > and now think the working set must always fit in memory. But this
> ! > is only one possible usecase, it is not the exclusive only one.
> !
> ! This is no-win situation as most of the complaints in recent years have been
> ! that Postgres was/is to conservative in its default settings and is not
> ! taking advantage of newer more powerful hardware.
>
> I know, and You got to the point; this is exactly what I am talking
> about: people take the abundance of ressources as granted.

Probably because the resources are there. My phone has computing power I
could only dream of when I was using a desktop of old.

>
> In Rel. 8 postgres was a breathtaking beauty of engineering: the style
> of old, driven to perfection.

I would hardly call version 8.x perfection, especially the attempt at
running on Windows natively.

> Now You're gradually sacrificing this, for the speed junkies and to
> protect those from mistakes who are not engineers.
>
> And no, I don't know how this could be solved: the more influential
> you get, the more driven you are by public demand; the less freedom
> you have to follow ideals.

Whose ideals? That is the issue. Postgres covers a broad spectrum of
uses and as such you will have complaints from either end that their
needs are not met.

>
> David Johnston thinks we must just not speak that out, we must instead
> behave like "the emperor's new clothes", and follow google's
> understanding of "positive values".
> Sorry, that doesn't work for me.

No, the complaint was that your pontificating interfered with your
problem description and got in the way of coming up with a solution.

--
Adrian Klaver
adrian(dot)klaver(at)aklaver(dot)com