Re: [Bug] Logical Replication failing if the DateStyle is different in Publisher & Subscriber

On Thu, 21 Oct 2021 at 22:46, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Japin Li <japinli(at)hotmail(dot)com> writes:
>> On Thu, 21 Oct 2021 at 19:54, Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
>>> BTW I think we can set the parameters from the subscriber side without
>>> additional network round trips by specifying the "options" parameter
>>> in the connection string, no?
>
>> Yes, we can. However, each client should be concerned the style for
>> datestyle, IMO it is boring.
>
> There's another issue here: the subscriber can run user-defined code
> (in triggers), while AFAIK the sender cannot.

Sorry, I'm not sure about this. Could you give me an example?

> People might be surprised
> if their triggers run with a datestyle setting different from the
> database's prevailing setting. So while I think it should be okay
> to set-and-forget the datestyle on the sender side, we could not get
> away with that in the subscriber. We'd have to set and unset for
> each row, much as (e.g.) postgres_fdw has to do.
>

Yeah! As Masahiko said, we can avoid the network round trips by specifying
the "options" parameter in the connection string.

If this way is more accepted, I'll update the patch later.

--
Regrads,
Japin Li.
ChengDu WenWu Information Technology Co.,Ltd.