On 03.04.25 10:07, Alvaro Herrera wrote: > The new flag is there for quick access by get_relation_info. We could > easily not have it otherwise, because clients don't need it, but its > lack would probably make planning measurably slower because it'd have to > do syscache access for every single not-null constraint to figure out if > it's valid or not.
In the v6 patch, you are adding a attnullability field to the CompactAttribute in the tuple descriptor and use that in get_relation_info(). That seems like the right approach, because then you're doing all that preprocessing about which constraint is active in the relcache. So I don't see where the extra pg_attribute field attnotnullvalid is getting used.'
attnotnullvalid is getting used to populate the CompatAttribute (populate_compact_attribute_internal).
The primary reason for adding a new field to pg_attribute is to avoid the need for an additional scan
of pg_constraint when populating CompatAttribute, as this extra scan introduces performance overhead
while retrieving catalog information for a relation.
