fixup! Check the partition_by clause for MVs during planning

Author: bkirwi

src/sql/src/plan/statement/ddl.rs

@@ -176,6 +176,9 @@ const MAX_NUM_COLUMNS: usize = 256;
 const MANAGED_REPLICA_PATTERN: std::sync::LazyLock<regex::Regex> =
     std::sync::LazyLock::new(|| regex::Regex::new(r"^r(\d)+$").unwrap());
 
+/// Given a relation desc and a column list, checks that:
+/// - the column list is a prefix of the desc;
+/// - all the listed columns are types that have meaningful Persist-level ordering.
 fn check_partition_by(desc: &RelationDesc, partition_by: &[ColumnName]) -> Result<(), PlanError> {
     for (idx, ((desc_name, desc_type), partition_name)) in
         desc.iter().zip(partition_by.iter()).enumerate()
@@ -184,7 +187,7 @@ fn check_partition_by(desc: &RelationDesc, partition_by: &[ColumnName]) -> Resul
             sql_bail!("PARTITION BY columns should be a prefix of the relation's columns (expected {desc_name} at index {idx}, got {partition_name})");
         }
         if !preserves_order(&desc_type.scalar_type) {
-            sql_bail!("PARTITION BY column {partition_name} has unsupported type",);
+            sql_bail!("PARTITION BY column {partition_name} has unsupported type");
         }
     }
     Ok(())