Thread

Commits

Doc: Fix oversight in commit 55cefadde8.
- 4441d6b2e449 19 (unreleased) landed
Doc: Add documentation for sequence synchronization.
- 55cefadde874 19 (unreleased) cited

DOC: pg_publication_rel.prrelid says sequences are possible

Peter Smith <smithpb2250@gmail.com> — 2026-04-01T04:59:54Z

Hi.

It is my understanding that only tables can exist in the
'pg_publication_rel' catalog.

e.g. only for these scenarios:
* included tables - when pg_publication.puballtables is false, these
are the specified FOR TABLE ... tables
* excluded tables - when pg_publication.puballtables is true, these
are the specified FOR ALL TABLES EXCEPT (TABLE ...) tables

In particular, AFAIK there is no way for sequences to be in the
'pg_publication_rel' catalog. e.g.
PG19 only supports "CREATE PUBLICATION ... FOR ALL SEQUENCES", which
is indicated by pg_publication.puballsequences.
PG19 has no support for "CREATE PUBLICATION ... FOR SEQUENCE seq1,seq2".

~~

Meanwhile, the DOCS page for 'pg_publication_rel' catalog [1] since
commit 55cefad says that sequences are also possible [2]:
------
prrelid oid (references pg_class.oid)
Reference to table or sequence
------

How are sequences possible? PSA a patch to correct it.

Or, am I mistaken here?

======
[1] https://www.postgresql.org/docs/devel/catalog-pg-publication-rel.html
[2] https://github.com/postgres/postgres/commit/55cefadde874e52b57f7b3c2232744e331f9d6bb#diff-686ae8747787a3c04698ca610350c2b7dc84618c5b28d4b90e94f6f8cd47bd8eR6571

Kind Regards,
Peter Smith.
Fujitsu Australia

Re: DOC: pg_publication_rel.prrelid says sequences are possible

shveta malik <shveta.malik@gmail.com> — 2026-04-01T07:05:34Z

On Wed, Apr 1, 2026 at 10:30 AM Peter Smith <smithpb2250@gmail.com> wrote:
>
> Hi.
>
> It is my understanding that only tables can exist in the
> 'pg_publication_rel' catalog.
>
> e.g. only for these scenarios:
> * included tables - when pg_publication.puballtables is false, these
> are the specified FOR TABLE ... tables
> * excluded tables - when pg_publication.puballtables is true, these
> are the specified FOR ALL TABLES EXCEPT (TABLE ...) tables
>
> In particular, AFAIK there is no way for sequences to be in the
> 'pg_publication_rel' catalog. e.g.
> PG19 only supports "CREATE PUBLICATION ... FOR ALL SEQUENCES", which
> is indicated by pg_publication.puballsequences.
> PG19 has no support for "CREATE PUBLICATION ... FOR SEQUENCE seq1,seq2".
>
> ~~
>
> Meanwhile, the DOCS page for 'pg_publication_rel' catalog [1] since
> commit 55cefad says that sequences are also possible [2]:
> ------
> prrelid oid (references pg_class.oid)
> Reference to table or sequence
> ------
>
> How are sequences possible? PSA a patch to correct it.
>
> Or, am I mistaken here?
>

You are right, pg_publication_rel can never have a seq entry while
pg_subscription_rel can have. The patch LGTM.

thanks
Shveta

Re: DOC: pg_publication_rel.prrelid says sequences are possible

Amit Kapila <amit.kapila16@gmail.com> — 2026-04-02T08:33:09Z

On Wed, Apr 1, 2026 at 12:35 PM shveta malik <shveta.malik@gmail.com> wrote:
>
> On Wed, Apr 1, 2026 at 10:30 AM Peter Smith <smithpb2250@gmail.com> wrote:
> >
> > Hi.
> >
> > It is my understanding that only tables can exist in the
> > 'pg_publication_rel' catalog.
> >
> > e.g. only for these scenarios:
> > * included tables - when pg_publication.puballtables is false, these
> > are the specified FOR TABLE ... tables
> > * excluded tables - when pg_publication.puballtables is true, these
> > are the specified FOR ALL TABLES EXCEPT (TABLE ...) tables
> >
> > In particular, AFAIK there is no way for sequences to be in the
> > 'pg_publication_rel' catalog. e.g.
> > PG19 only supports "CREATE PUBLICATION ... FOR ALL SEQUENCES", which
> > is indicated by pg_publication.puballsequences.
> > PG19 has no support for "CREATE PUBLICATION ... FOR SEQUENCE seq1,seq2".
> >
> > ~~
> >
> > Meanwhile, the DOCS page for 'pg_publication_rel' catalog [1] since
> > commit 55cefad says that sequences are also possible [2]:
> > ------
> > prrelid oid (references pg_class.oid)
> > Reference to table or sequence
> > ------
> >
> > How are sequences possible? PSA a patch to correct it.
> >
> > Or, am I mistaken here?
> >
>
> You are right, pg_publication_rel can never have a seq entry while
> pg_subscription_rel can have. The patch LGTM.
>

Pushed.

-- 
With Regards,
Amit Kapila.

Re: DOC: pg_publication_rel.prrelid says sequences are possible

Peter Smith <smithpb2250@gmail.com> — 2026-04-07T00:25:05Z

On Thu, Apr 2, 2026 at 7:33 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> On Wed, Apr 1, 2026 at 12:35 PM shveta malik <shveta.malik@gmail.com> wrote:
> >
...
> > You are right, pg_publication_rel can never have a seq entry while
> > pg_subscription_rel can have. The patch LGTM.
> >
>
> Pushed.

Thanks for pushing.

======
Kind Regards,
Peter Smith.
Fujitsu Australia