Re: not null constraints, again

On 2024-Nov-08, jian he wrote:

> > Here's v11, which I intended to commit today, but didn't get around to.
> > CI is happy with it, so I'll probably do it tomorrow first thing.
> >
> v11 still has column_constraint versus table_constraint inconsistency.
>
> create table t7 (a int generated by default as identity, constraint
> foo not null a no inherit, b int);
> create table t7 (a int generated by default as identity not null no
> inherit, b int);
> create table t8 (a serial, constraint foo1 not null a no inherit);
> create table t8 (a serial not null no inherit, b int);
>
> i solved this issue at [1],

Ah yeah, that stuff. Your commit message said it was a refactoring so I
hadn't paid too much attention to it, but it's in fact not a refactoring
at all. I included it with a large comment explaining why we do it that
way and that we may want to remove it in the future. I also included
these four sentences above in the tests, and pushed it after checking
that the CI results are clean.

Yesterday I verified that pg_upgrade works with the regression database
from 12 onwards. I know the buildfarm uses a different way to do the
pg_upgrade test, so there's no way to know if it'll work ahead of time.

But we'll see what else the buildfarm has to say now that I pushed it ...

--
Álvaro Herrera Breisgau, Deutschland — https://www.EnterpriseDB.com/