On Monday, April 4, 2022 10:47:51 PM CEST David Rowley wrote: > On Tue, 5 Apr 2022 at 01:21, J. Roeleveld <joost@xxxxxxxxxxxx> wrote: > > Personally, I think NULL should be treated as a seperate value and not > > lead to strange behaviour. > > I think the rationale behind IN and NOT IN are that c IN(1,2,3) is > equivalent of writing: c = 1 OR c = 2 OR c = 3, whereas NOT IN(1,2,3) > would be the same as c <> 1 AND c <> 2 AND c <> 3. You can imagine > what would happen in the latter case if you replaced 3 with NULL. "c > <> NULL" is NULL therefore, due to the quals being ANDed, will cause > the WHERE clause not to match anything. > > In any case, it's what the SQL standard says, so that's the way we do it. I agree with following the standard. If I would feel really strongly about this (I don't), it would be up to me to try and convince others. And I have got better things to do with my time. :) -- Joost
