Re: Proposal: Conflict log history table for Logical Replication

On Sun, Aug 23, 2026 at 1:45 PM Zhijie Hou (Fujitsu)
<houzj(dot)fnst(at)fujitsu(dot)com> wrote:
>
> On Friday, August 21, 2026 6:46 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> >
> > On Thu, Aug 20, 2026 at 8:15 AM Zhijie Hou (Fujitsu) <houzj(dot)fnst(at)fujitsu(dot)com>
> > wrote:
> > >
> > > One more thing 0001 could potentially do is replace the existing
> > > references to stream_xid with ApplyRemoteCtx.remote_xid. It's not done
> > > in the patch yet, but can be if we reach consensus.
> > >
> >
> > We can do this without breaking anything but this doesn't sound
> > fundamentally correct. Because stream_xid is valid only inside a STREAM
> > START/STOP block whereas remote_ctx.remote_xid is valid for the whole
> > transaction(including the COMMIT/PREPARE/ABORT phase), and is used for
> > non-streamed transactions. So remote_xid is deliberately valid at moments
> > when stream_xid is deliberately Invalid. So, I suggest leaving this as it is.
> >
> > I have made few changes in 0001 refactor patch (a) Change the mix use of
> > errarg and remote_ctx in apply_error_callback as that looks confusing, instead
> > used a shorter variable name ctx so that some of the lines where it is used
> > doesn't overflow, (b) changed comments at various places. Let me know what
> > you think of attached.
>
> Thanks for updating the patch, it looks good to me.
>

Pushed.

--
With Regards,
Amit Kapila.