1. Semantic design change 1.1 Naming problem In 7X, partitions are first-class tables. There is no such thing as alias (partition name) in 7X, therefore pg_partitons.partitionname and pg_partitons.parentpartitionname no longer make sense and not available either. Take the following table partrl as an example: -- Classic Syntax -- Multi-level range-list partitioned table CREATE TABLE partrl (a int, b int, c int) DISTRIBUTED BY (a)

Hello hackers, I've been playing with partitioning on GPDB7 with mixed flavors of grammar: upstream Postgres grammar and legacy GPDB grammar. I have the following findings and I'm looking for your thoughts. Context Historically[1] Greenplum didn't allow a partitioned table to have zero child. This guard exists in gpdb6 and all prior versions for CREATE TABLE and ALTER TABLE DROP PARTITION.

A brief intro on Share Input Scan -- setup drop table if exists foo; drop table if exists bar; drop table if exists jazz; CREATE TABLE foo (a int, b int); CREATE TABLE bar (c int, d int); CREATE TABLE jazz(e int, f int);

PR: https://github.com/greenplum-db/gpdb/pull/12550 Repro -- setup tables drop table if exists t1; create table t1 (a int, b int); drop table if exists t2; create table t2 (a int);