diff --git a/src/backend/executor/execExpr.c b/src/backend/executor/execExpr.c
index 9653652f6d..33f5e3a8fe 100644
--- a/src/backend/executor/execExpr.c
+++ b/src/backend/executor/execExpr.c
@@ -1631,16 +1631,16 @@ ExecInitExprRec(Expr *node, ExprState *state,
 				{
 					/* generic record, use types of given expressions */
 					tupdesc = ExecTypeFromExprList(rowexpr->args);
+					/* ... but adopt RowExpr's column aliases */
+					ExecTypeSetColNames(tupdesc, rowexpr->colnames);
+					/* Bless the tupdesc so it can be looked up later */
+					BlessTupleDesc(tupdesc);
 				}
 				else
 				{
 					/* it's been cast to a named type, use that */
 					tupdesc = lookup_rowtype_tupdesc_copy(rowexpr->row_typeid, -1);
 				}
-				/* In either case, adopt RowExpr's column aliases */
-				ExecTypeSetColNames(tupdesc, rowexpr->colnames);
-				/* Bless the tupdesc in case it's now of type RECORD */
-				BlessTupleDesc(tupdesc);
 
 				/*
 				 * In the named-type case, the tupdesc could have more columns
diff --git a/src/backend/executor/execExprInterp.c b/src/backend/executor/execExprInterp.c
index 4524034d5d..07b33531a6 100644
--- a/src/backend/executor/execExprInterp.c
+++ b/src/backend/executor/execExprInterp.c
@@ -3991,12 +3991,8 @@ ExecEvalWholeRowVar(ExprState *state, ExprEvalStep *op, ExprContext *econtext)
 			 * generates an INT4 NULL regardless of the dropped column type).
 			 * If we find a dropped column and cannot verify that case (1)
 			 * holds, we have to use the slow path to check (2) for each row.
-			 *
-			 * If vartype is a domain over composite, just look through that
-			 * to the base composite type.
 			 */
-			var_tupdesc = lookup_rowtype_tupdesc_domain(variable->vartype,
-														-1, false);
+			var_tupdesc = lookup_rowtype_tupdesc(variable->vartype, -1);
 
 			slot_tupdesc = slot->tts_tupleDescriptor;
 
@@ -4033,9 +4029,8 @@ ExecEvalWholeRowVar(ExprState *state, ExprEvalStep *op, ExprContext *econtext)
 
 			/*
 			 * Use the variable's declared rowtype as the descriptor for the
-			 * output values, modulo possibly assigning new column names
-			 * below. In particular, we *must* absorb any attisdropped
-			 * markings.
+			 * output values.  In particular, we *must* absorb any
+			 * attisdropped markings.
 			 */
 			oldcontext = MemoryContextSwitchTo(econtext->ecxt_per_query_memory);
 			output_tupdesc = CreateTupleDescCopy(var_tupdesc);
@@ -4053,39 +4048,38 @@ ExecEvalWholeRowVar(ExprState *state, ExprEvalStep *op, ExprContext *econtext)
 			oldcontext = MemoryContextSwitchTo(econtext->ecxt_per_query_memory);
 			output_tupdesc = CreateTupleDescCopy(slot->tts_tupleDescriptor);
 			MemoryContextSwitchTo(oldcontext);
-		}
 
-		/*
-		 * Construct a tuple descriptor for the composite values we'll
-		 * produce, and make sure its record type is "blessed".  The main
-		 * reason to do this is to be sure that operations such as
-		 * row_to_json() will see the desired column names when they look up
-		 * the descriptor from the type information embedded in the composite
-		 * values.
-		 *
-		 * We already got the correct physical datatype info above, but now we
-		 * should try to find the source RTE and adopt its column aliases, in
-		 * case they are different from the original rowtype's names.  For
-		 * example, in "SELECT foo(t) FROM tab t(x,y)", the first two columns
-		 * in the composite output should be named "x" and "y" regardless of
-		 * tab's column names.
-		 *
-		 * If we can't locate the RTE, assume the column names we've got are
-		 * OK.  (As of this writing, the only cases where we can't locate the
-		 * RTE are in execution of trigger WHEN clauses, and then the Var will
-		 * have the trigger's relation's rowtype, so its names are fine.)
-		 * Also, if the creator of the RTE didn't bother to fill in an eref
-		 * field, assume our column names are OK.  (This happens in COPY, and
-		 * perhaps other places.)
-		 */
-		if (econtext->ecxt_estate &&
-			variable->varno <= econtext->ecxt_estate->es_range_table_size)
-		{
-			RangeTblEntry *rte = exec_rt_fetch(variable->varno,
-											   econtext->ecxt_estate);
+			/*
+			 * It's possible that the input slot is a relation scan slot and
+			 * so is marked with that relation's rowtype.  But we're supposed
+			 * to be returning RECORD, so reset to that.
+			 */
+			output_tupdesc->tdtypeid = RECORDOID;
+			output_tupdesc->tdtypmod = -1;
 
-			if (rte->eref)
-				ExecTypeSetColNames(output_tupdesc, rte->eref->colnames);
+			/*
+			 * We already got the correct physical datatype info above, but
+			 * now we should try to find the source RTE and adopt its column
+			 * aliases, since it's unlikely that the input slot has the
+			 * desired names.
+			 *
+			 * If we can't locate the RTE, assume the column names we've got
+			 * are OK.  (As of this writing, the only cases where we can't
+			 * locate the RTE are in execution of trigger WHEN clauses, and
+			 * then the Var will have the trigger's relation's rowtype, so its
+			 * names are fine.)  Also, if the creator of the RTE didn't bother
+			 * to fill in an eref field, assume our column names are OK. (This
+			 * happens in COPY, and perhaps other places.)
+			 */
+			if (econtext->ecxt_estate &&
+				variable->varno <= econtext->ecxt_estate->es_range_table_size)
+			{
+				RangeTblEntry *rte = exec_rt_fetch(variable->varno,
+												   econtext->ecxt_estate);
+
+				if (rte->eref)
+					ExecTypeSetColNames(output_tupdesc, rte->eref->colnames);
+			}
 		}
 
 		/* Bless the tupdesc if needed, and save it in the execution state */
diff --git a/src/backend/executor/execTuples.c b/src/backend/executor/execTuples.c
index c381f3465c..b4b8f2eeb2 100644
--- a/src/backend/executor/execTuples.c
+++ b/src/backend/executor/execTuples.c
@@ -2022,51 +2022,40 @@ ExecTypeFromExprList(List *exprList)
 }
 
 /*
- * ExecTypeSetColNames - set column names in a TupleDesc
+ * ExecTypeSetColNames - set column names in a RECORD TupleDesc
  *
  * Column names must be provided as an alias list (list of String nodes).
- *
- * For some callers, the supplied tupdesc has a named rowtype (not RECORD)
- * and it is moderately likely that the alias list matches the column names
- * already present in the tupdesc.  If we do change any column names then
- * we must reset the tupdesc's type to anonymous RECORD; but we avoid doing
- * so if no names change.
  */
 void
 ExecTypeSetColNames(TupleDesc typeInfo, List *namesList)
 {
-	bool		modified = false;
 	int			colno = 0;
 	ListCell   *lc;
 
+	/* It's only OK to change col names in a not-yet-blessed RECORD type */
+	Assert(typeInfo->tdtypeid == RECORDOID);
+	Assert(typeInfo->tdtypmod < 0);
+
 	foreach(lc, namesList)
 	{
 		char	   *cname = strVal(lfirst(lc));
 		Form_pg_attribute attr;
 
-		/* Guard against too-long names list */
+		/* Guard against too-long names list (probably can't happen) */
 		if (colno >= typeInfo->natts)
 			break;
 		attr = TupleDescAttr(typeInfo, colno);
 		colno++;
 
-		/* Ignore empty aliases (these must be for dropped columns) */
-		if (cname[0] == '\0')
+		/*
+		 * Do nothing for empty aliases or dropped columns (these cases
+		 * probably can't arise in RECORD types, either)
+		 */
+		if (cname[0] == '\0' || attr->attisdropped)
 			continue;
 
-		/* Change tupdesc only if alias is actually different */
-		if (strcmp(cname, NameStr(attr->attname)) != 0)
-		{
-			namestrcpy(&(attr->attname), cname);
-			modified = true;
-		}
-	}
-
-	/* If we modified the tupdesc, it's now a new record type */
-	if (modified)
-	{
-		typeInfo->tdtypeid = RECORDOID;
-		typeInfo->tdtypmod = -1;
+		/* OK, assign the column name */
+		namestrcpy(&(attr->attname), cname);
 	}
 }
 
diff --git a/src/test/regress/expected/rowtypes.out b/src/test/regress/expected/rowtypes.out
index 2a273f8404..ec7f17ec21 100644
--- a/src/test/regress/expected/rowtypes.out
+++ b/src/test/regress/expected/rowtypes.out
@@ -1004,18 +1004,8 @@ select row_to_json(i) from int8_tbl i;
  {"q1":4567890123456789,"q2":-4567890123456789}
 (5 rows)
 
+-- since "i" is of type "int8_tbl", attaching aliases doesn't change anything:
 select row_to_json(i) from int8_tbl i(x,y);
-                 row_to_json                  
-----------------------------------------------
- {"x":123,"y":456}
- {"x":123,"y":4567890123456789}
- {"x":4567890123456789,"y":123}
- {"x":4567890123456789,"y":4567890123456789}
- {"x":4567890123456789,"y":-4567890123456789}
-(5 rows)
-
-create temp view vv1 as select * from int8_tbl;
-select row_to_json(i) from vv1 i;
                   row_to_json                   
 ------------------------------------------------
  {"q1":123,"q2":456}
@@ -1025,16 +1015,7 @@ select row_to_json(i) from vv1 i;
  {"q1":4567890123456789,"q2":-4567890123456789}
 (5 rows)
 
-select row_to_json(i) from vv1 i(x,y);
-                 row_to_json                  
-----------------------------------------------
- {"x":123,"y":456}
- {"x":123,"y":4567890123456789}
- {"x":4567890123456789,"y":123}
- {"x":4567890123456789,"y":4567890123456789}
- {"x":4567890123456789,"y":-4567890123456789}
-(5 rows)
-
+-- in these examples, we'll report the exposed column names of the subselect:
 select row_to_json(ss) from
   (select q1, q2 from int8_tbl) as ss;
                   row_to_json                   
diff --git a/src/test/regress/sql/rowtypes.sql b/src/test/regress/sql/rowtypes.sql
index 83cf4a1e79..3c83727f81 100644
--- a/src/test/regress/sql/rowtypes.sql
+++ b/src/test/regress/sql/rowtypes.sql
@@ -413,12 +413,10 @@ select longname(f) from fullname f;
 --
 
 select row_to_json(i) from int8_tbl i;
+-- since "i" is of type "int8_tbl", attaching aliases doesn't change anything:
 select row_to_json(i) from int8_tbl i(x,y);
 
-create temp view vv1 as select * from int8_tbl;
-select row_to_json(i) from vv1 i;
-select row_to_json(i) from vv1 i(x,y);
-
+-- in these examples, we'll report the exposed column names of the subselect:
 select row_to_json(ss) from
   (select q1, q2 from int8_tbl) as ss;
 select row_to_json(ss) from