diff --git a/src/backend/optimizer/prep/prepjointree.c b/src/backend/optimizer/prep/prepjointree.c
index 282589dec81..74823e8437a 100644
--- a/src/backend/optimizer/prep/prepjointree.c
+++ b/src/backend/optimizer/prep/prepjointree.c
@@ -2279,8 +2279,8 @@ pullup_replace_vars_callback(Var *var,
 		 * If generating an expansion for a var of a named rowtype (ie, this
 		 * is a plain relation RTE), then we must include dummy items for
 		 * dropped columns.  If the var is RECORD (ie, this is a JOIN), then
-		 * omit dropped columns. Either way, attach column names to the
-		 * RowExpr for use of ruleutils.c.
+		 * omit dropped columns.  In the latter case, attach column names to
+		 * the RowExpr for use of the executor and ruleutils.c.
 		 *
 		 * In order to be able to cache the results, we always generate the
 		 * expansion with varlevelsup = 0, and then adjust if needed.
@@ -2301,7 +2301,7 @@ pullup_replace_vars_callback(Var *var,
 		rowexpr->args = fields;
 		rowexpr->row_typeid = var->vartype;
 		rowexpr->row_format = COERCE_IMPLICIT_CAST;
-		rowexpr->colnames = colnames;
+		rowexpr->colnames = (var->vartype == RECORDOID) ? colnames : NIL;
 		rowexpr->location = var->location;
 		newnode = (Node *) rowexpr;
 
diff --git a/src/backend/optimizer/util/var.c b/src/backend/optimizer/util/var.c
index a0543b7f47b..2b44ef3e176 100644
--- a/src/backend/optimizer/util/var.c
+++ b/src/backend/optimizer/util/var.c
@@ -809,6 +809,7 @@ flatten_join_alias_vars_mutator(Node *node,
 			rowexpr->args = fields;
 			rowexpr->row_typeid = var->vartype;
 			rowexpr->row_format = COERCE_IMPLICIT_CAST;
+			/* vartype will always be RECORDOID, so we always need colnames */
 			rowexpr->colnames = colnames;
 			rowexpr->location = var->location;
 
diff --git a/src/backend/rewrite/rewriteManip.c b/src/backend/rewrite/rewriteManip.c
index 708e5453e31..101c39553ae 100644
--- a/src/backend/rewrite/rewriteManip.c
+++ b/src/backend/rewrite/rewriteManip.c
@@ -1424,8 +1424,8 @@ ReplaceVarsFromTargetList_callback(Var *var,
 		 * If generating an expansion for a var of a named rowtype (ie, this
 		 * is a plain relation RTE), then we must include dummy items for
 		 * dropped columns.  If the var is RECORD (ie, this is a JOIN), then
-		 * omit dropped columns.  Either way, attach column names to the
-		 * RowExpr for use of ruleutils.c.
+		 * omit dropped columns.  In the latter case, attach column names to
+		 * the RowExpr for use of the executor and ruleutils.c.
 		 */
 		expandRTE(rcon->target_rte,
 				  var->varno, var->varlevelsup, var->location,
@@ -1438,7 +1438,7 @@ ReplaceVarsFromTargetList_callback(Var *var,
 		rowexpr->args = fields;
 		rowexpr->row_typeid = var->vartype;
 		rowexpr->row_format = COERCE_IMPLICIT_CAST;
-		rowexpr->colnames = colnames;
+		rowexpr->colnames = (var->vartype == RECORDOID) ? colnames : NIL;
 		rowexpr->location = var->location;
 
 		return (Node *) rowexpr;
diff --git a/src/include/nodes/primnodes.h b/src/include/nodes/primnodes.h
index dab5c4ff5de..439e4b4a9db 100644
--- a/src/include/nodes/primnodes.h
+++ b/src/include/nodes/primnodes.h
@@ -1052,15 +1052,13 @@ typedef struct ArrayExpr
  * than vice versa.)  It is important not to assume that length(args) is
  * the same as the number of columns logically present in the rowtype.
  *
- * colnames provides field names in cases where the names can't easily be
- * obtained otherwise.  Names *must* be provided if row_typeid is RECORDOID.
- * If row_typeid identifies a known composite type, colnames can be NIL to
- * indicate the type's cataloged field names apply.  Note that colnames can
- * be non-NIL even for a composite type, and typically is when the RowExpr
- * was created by expanding a whole-row Var.  This is so that we can retain
- * the column alias names of the RTE that the Var referenced (which would
- * otherwise be very difficult to extract from the parsetree).  Like the
- * args list, colnames is one-for-one with physical fields of the rowtype.
+ * colnames provides field names if the ROW() result is of type RECORD.
+ * Names *must* be provided if row_typeid is RECORDOID; but if it is a
+ * named composite type, colnames will be ignored in favor of using the
+ * type's cataloged field names, so colnames should be NIL.  Like the
+ * args list, colnames is defined to be one-for-one with physical fields
+ * of the rowtype (although dropped columns shouldn't appear in the
+ * RECORD case, so this fine point is currently moot).
  */
 typedef struct RowExpr
 {