Re: pg_index.indisreplident and invalid indexes

On Fri, Aug 28, 2020 at 10:15:37AM +0200, Dmitry Dolgov wrote:
> Thanks for the patch. It sounds right, so no objections from me. But I
> wonder if something similar has to be done also for
> index_concurrently_swap function?

As of index.c, this already happens:
    /* Preserve indisreplident in the new index */
    newIndexForm->indisreplident = oldIndexForm->indisreplident;
    oldIndexForm->indisreplident = false;

In short, the new concurrent index is created first with
indisreplident = false, and when swapping the old and new indexes, the
new index inherits the setting of the old one, and the old one planned
for drop uses indisreplident = false when swapping.
--
Michael