Re: BUG #17245: Index corruption involving deduplicated entries

On 2021-Oct-29, Peter Geoghegan wrote:

> However, this relation is a B-Tree index, where TIDs/page offset
> numbers are not stable at all.
> 
> Almost all individual index tuple inserts onto a B-Tree page put the
> new index tuple "between" existing index tuples. This will "shift"
> whatever index tuples are to the right of the position of the new
> tuple.

Bah, of course.  Sorry for the noise.

-- 
Álvaro Herrera              Valdivia, Chile  —  https://www.EnterpriseDB.com/
"Puedes vivir sólo una vez, pero si lo haces bien, una vez es suficiente"