"Pavan Deolasee" <pavan.deolasee@enterprisedb.com> writes:
> If that assumption is true (and it must be true for us to move
> the chain in pieces), doesn't that mean we don't really need to
> move the RECENTLY_DEAD tuples preceding a DEAD tuple ?
As I've already said several times: they are dead, but at least for
VACUUM FULL's purposes it seems unreasonably difficult to determine that
and remove them. The point at which we'd figure this out is after we've
already performed dead-tuple removal (at least for some of the pages
involved). It would make for a significant increase in logical
complexity, and that code is too damn complicated already. Since we
know this is a seldom-seen corner case, I'm not going to risk
introducing new bugs to recycle a few tuples a bit sooner.
regards, tom lane