Re: Handle infinite recursion in logical replication setup

On Tue, Aug 8, 2023 at 6:52 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>
> On Tue, Aug 8, 2023 at 1:50 PM Peter Eisentraut <peter(at)eisentraut(dot)org> wrote:
> >
> > This patch added the following error message:
> >
> > errdetail_plural("Subscribed publication %s is subscribing to other
> > publications.",
> > "Subscribed publications %s are subscribing to other publications.",
> > list_length(publist), pubnames->data),
> >
> > But in PostgreSQL, a publication cannot subscribe to a publication, so
> > this is not giving accurate information. Apparently, what it is trying
> > to say is that
> >
> > The subscription that you are creating subscribes to publications that
> > contain tables that are written to by other subscriptions.
> >
> > Can we get to a more accurate wording like this?
> >
>
> +1 for changing the message as per your suggestion.
>

PSA a patch to change this message text. The message now has wording
similar to the suggestion.

> > There is also a translatability issue there, in the way the publication
> > list is pasted into the message.
> >

The name/list substitution is now done within parentheses, which AFAIK
will be enough to eliminate any translation ambiguities.

> > Is the list of affected publications really that interesting? I wonder
> > whether the list of affected tables might be more relevant?
> >
>
> In that case, we need to specify both schema name and table name in
> that case. I guess the list could be very long and not sure what to do
> for schema publications ( Create Publication ... For Schema).

Right, IIUC that was the reason for not choosing to show the tables in
the original patch -- i.e. the list might easily be very long with
100s or 1000s of tables it, and so inappropriate to substitute in the
message. OTOH, showing only problematic publications is a short list
but it is still more useful than showing nothing (e.g. there other
publications of the subscription might be OK so those ones are not
listed)

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