Bruce Momjian wrote:
>
> So, yes, I suppose there is an inconsistency there. Zero-sized
> tables(according to vacuum), use index, while tables with some data
> don't use index.
>
> How does the system know there is a row in there if you didn't run
> vacuum? That confuses me.
Create index updates ntuples & npages in pg_class...
Vadim
