On 2020-04-08 07:45, Amit Langote wrote:
On Wed, Apr 8, 2020 at 1:22 AM Amit Langote <amitlangot...@gmail.com> wrote:
On Tue, Apr 7, 2020 at 6:01 PM Peter Eisentraut
<peter.eisentr...@2ndquadrant.com> wrote:
The descriptions of the new fields in RelationSyncEntry don't seem to
match the code accurately, or at least it's confusing.
replicate_as_relid is always filled in with an ancestor, even if
pubviaroot is not set.
Given this confusion, I have changed how replicate_as_relid works so
that it's now always set -- if different from the relation's own OID,
the code for "publishing via root" kicks in in various places.
I think the pubviaroot field is actually not necessary. We only need
replicate_as_relid.
Looking through the code, I agree. I guess I only kept it around to
go with pubupdate, etc.
Think I broke truncate replication with this. Fixed in the attached
updated patch.
All committed.
Thank you and everyone very much for working on this. I'm very happy
that these two features from PG10 have finally met. :)
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
