Re: [HACKERS] VACUUM and ANALYZE disagreeing on what reltuples means

I wrote:
> Tomas Vondra <tomas.vondra@2ndquadrant.com> writes:
>> The 0002 part is the main part, unifying the definition of reltuples on
>> three main places:

> On to this part ...

I've pushed 0002 with several corrections: it did not seem to me that
you'd correctly propagated what ANALYZE is doing into CREATE INDEX or
pgstattuple.  Also, since one of the things VACUUM does with num_tuples
is to report it as "the total number of non-removable tuples", I thought
we'd better leave that calculation alone.  I made the added code count
live tuples in a new variable live_tuples, instead.

            regards, tom lane