RE: Fix pg_publication_tables to exclude generated columns

On Wed, Jan 11, 2023 2:40 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>
> On Wed, Jan 11, 2023 at 10:07 AM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> >
> > Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> writes:
> > >> On Mon, Jan 9, 2023 11:06 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > >>> We could just not fix it in the back branches. I'd argue that this is
> > >>> as much a definition change as a bug fix, so it doesn't really feel
> > >>> like something to back-patch anyway.
> >
> > > So, if we don't backpatch then it could lead to an error when it
> > > shouldn't have which is clearly a bug. I think we should backpatch
> > > this unless Tom or others are against it.
> >
> > This isn't a hill that I'm ready to die on ... but do we have any field
> > complaints about this? If not, I still lean against a back-patch.
> > I think there's a significant risk of breaking case A while fixing
> > case B when we change this behavior, and that's something that's
> > better done only in a major release.
> >
>
> Fair enough, but note that there is a somewhat related problem for
> dropped columns [1] as well. While reviewing that it occurred to me
> that generated columns also have a similar problem which leads to this
> thread (it would have been better if there is a mention of the same in
> the initial email). Now, as symptoms are similar, I think we shouldn't
> back-patch that as well, otherwise, it will appear to be partially
> fixed. What do you think?
>
> [1] - https://www.postgresql.org/message-
> id/OSZPR01MB631087C65BA81E1FEE5A60D2FDF59%40OSZPR01MB6310.jpnpr
> d01.prod.outlook.com
>

I agree to only fix them on HEAD.

I merged this patch and the one in [1] as they are similar problems. Please
see the attached patch.

I removed the changes in tablesync.c which simplified the query in
fetch_remote_table_info(), because it only works for publishers of v16. Those
changes are based on pg_get_publication_tables() returning all columns when no
column list is specified, but publishers of v15 return NULL in that case.

[1] https://www.postgresql.org/message-id/OSZPR01MB631087C65BA81E1FEE5A60D2FDF59%40OSZPR01MB6310.jpnprd01.prod.outlook.com

Regards,
Shi yu