Re: Is the Halloween problem an issue in Postgres

On Wed, Dec 02, 2020 at 11:02:07PM +0100, Thomas Kellerer <shammat(at)gmx(dot)net> wrote:

> guyren(at)icloud(dot)com schrieb am 02.12.2020 um 21:27:
> > The Halloween problem is that it is a challenge for the database if
> > you’re updating a field that is also in the WHERE clause of the same
> > query.
> >
> > I just saw a presentation from someone about how in SQL Server he
> > recommended writing changes to a temp table and then writing them to
> > the table as being much more efficient.
>
> It sounds strange to me, that this _is_ actually a problem.
>
> Why exactly is that a problem in SQL Server?
> And what are the consequences if you do it nevertheless.

According to wikipedia, this problem was discovered on
Halloween day, 1976. I find it hard to believe that any
database would still exhibit that behaviour 44 years
later.

cheers,
raf