On Fri, Jan 12, 2018 at 5:06 PM, Andres Freund <and...@anarazel.de> wrote: > OTOH, it seems quite likely that we'll add more transaction-lifetime > shared data (e.g. combocid), so building per-xact infrastructure > actually seems like a good idea.
Sure, but there's no urgency about it. Particularly if the first cut at this is implemented using dshash, moving the dshash to a different DSA with a different lifespan later should be easy. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company