RE: Move global variables of pgoutput to plugin private scope.

On Wednesday, September 27, 2023 12:45 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com>
>
> On Wed, Sep 27, 2023 at 9:46 AM Michael Paquier <michael(at)paquier(dot)xyz>
> wrote:
> >
> > On Wed, Sep 27, 2023 at 09:39:19AM +0530, Amit Kapila wrote:
> > > On Wed, Sep 27, 2023 at 9:10 AM Michael Paquier <michael(at)paquier(dot)xyz>
> wrote:
> > >> Err, actually, I am going to disagree here for the patch of HEAD.
> > >> It seems to me that there is zero need for pgoutput.h and we don't
> > >> need to show PGOutputData to the world. The structure is internal
> > >> to Pgoutput.c and used only by its internal static routines.
> > >
> > > Do you disagree with the approach for the PG16 patch or HEAD? You
> > > mentioned HEAD but your argument sounds like you disagree with a
> > > different approach for PG16.
> >
> > Only HEAD where the structure should be moved from pgoutput.h to
> > pgoutput.c, IMO.
> >
>
> It's like that from the beginning. Now, even if we want to move, your
> suggestion is not directly related to this patch as we are just changing one field,
> and that too to fix a bug. We should start a separate thread to gather a broader
> consensus if we want to move the exposed structure to an internal file.

While searching the code, I noticed one postgres fork where the PGoutputData is
used in other places, although it's a separate fork, but it seems better to
discuss the removal separately.

[1] https://github.com/Tencent/TBase/blob/7cf7f8afbcab7290538ad5e65893561710be3dfa/src/backend/replication/walsender.c#L100

Best Regards,
Hou zj