On Fri, Aug 27, 2021 at 3:31 AM Peter Smith <smithpb2250@gmail.com> wrote:
>
> On Thu, Aug 26, 2021 at 9:13 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
> >
> > On Thu, Aug 26, 2021 at 3:41 PM Peter Smith <smithpb2250@gmail.com> wrote:
> >
> > Apart from Truncate, it will also be a waste if any error happens
> > before actually evaluating the filter, tomorrow there could be other
> > operations like replication of sequences (I have checked that proposed
> > patch for sequences uses get_rel_sync_entry) where we don't need to
> > build ExprState (as filters might or might not be there). So, it would
> > be better to avoid cache lookups in those cases if possible. I still
> > think doing expensive things like preparing expressions should ideally
> > be done only when it is required.
>
> OK. Per your suggestion, I will try to move as much of the row-filter
> cache code as possible out of the get_rel_sync_entry function and into
> the pgoutput_row_filter function.
>
I could think of more scenarios where doing this work in
get_rel_sync_entry() could cost us without any actual need for it.
Consider, the user has published only 'update' and 'delete' operation
for a publication, then in the system there are inserts followed
truncate or any ddl which generates invalidation, for such a case, for
each change we need to rebuild the row_filters but we won't use it.
Similarly, this can happen in any other combination of DML and DDL
operations where the DML operation is not published. I don't want to
say that this is the most common scenario but it is important to do
expensive work when it is actually required, otherwise, there could be
cases where it might hit us.
--
With Regards,
Amit Kapila.