On Fri, Sep 3, 2021 at 4:06 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> On Wed, Sep 1, 2021 at 12:05 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
> >
> > On Wed, Sep 1, 2021 at 8:52 AM Greg Nancarrow <gregn4422@gmail.com> wrote:
> > >
> >
> > > I'd expect a lot of users to naturally think that "ALTER PUBLICATION
> > > pub1 DROP ALL TABLES IN SCHEMA sc1;" would drop from the publication
> > > all tables that are in schema "sc1", which is not what it is currently
> > > doing.
> > > Since the syntax was changed to specifically refer to FOR ALL TABLES
> > > IN SCHEMA rather than FOR SCHEMA, then now it's clear we're referring
> > > to tables only, when specifying "... FOR ALL TABLES in sc1, TABLE
> > > sc1.test", so IMHO it's reasonable to remove duplicates here, rather
> > > than treating these as somehow separate ways of referencing the same
> > > table.
> > >
> >
> > I see your point and if we decide to go this path then it is better to
> > give an error than silently removing duplicates.
> >
>
> Today, I have thought about this point again and it seems better to
> give an error in this case and let the user take the action rather
> than silently removing such tables to avoid any confusion.
I have handled this to throw an error. This is handled as part of v25
patch attached at [1].
[1] - https://www.postgresql.org/message-id/CALDaNm2SytXy2TDnzzYkXWKgNp74ssPBXrkMXEyac1qVYSRkbw%40mail.gmail.com
Regards,
Vignesh