RE: Add support for specifying tables in pg_createsubscriber.

On Saturday, August 2, 2025 12:59 AM Andrew Dunstan <andrew@dunslane.net>  wrote:
> On 2025-08-01 Fr 11:03 AM, Zhijie Hou (Fujitsu) wrote:
> > On Friday, August 1, 2025 8:56 PM Andrew Dunstan mailto:andrew@dunslane.net
> wrote:
> 
> > > > We have another example to consider: pg_amcheck, which allows users to
> > > > specify multiple databases. 
> > > 
> > > I don't think that's quite the point, as I understand it. pg_amcheck might
> > > allow you to have multiple --database arguments, but I don't think it depends
> > > on the order of arguments. You didn't answer his question about what
> > > getopt_long() does. I don't recall if it is free to mangle the argument order.
> > 
> > I think you might misunderstand my proposal. I am suggesting an alternative
> > interface style that employs database-qualified table names, which doesn't
> > depend on the order of options. This style is already used by pg_amcheck when
> > dealing with multiple database specifications. I referenced pg_amcheck as an
> > example. 
> 
> I simple took your own description: The attached patch introduces a new
> '--table' option that can be specified after each '--database' argument. Maybe I
> need some remedial English, but to me that "after" says that argument order is
> significant.

Allow me to clarify the situation. The description you referenced is the
original interface proposed by the author in the initial email. However, it was
found to be unstable due to its reliance on the argument order. In response to
the discussion, instead of supporting the original interface, I suggested an
alternative interface to consider, which is the one that does not depend on
argument order, as I mentioned in my previous email.

Best Regards,
Hou zj