diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 8fb28f491db..52489a2e90d 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -183,8 +183,9 @@ should_refetch_tuple(TM_Result res, TM_FailureData *tmfd)
  * 'skipduplicates' specifies whether the first matching tuple can be used
  * without comparing it against 'searchslot', or at worst by comparing just
  * the index key columns (when the index scan requires a recheck, as with a
- * lossy GiST index for WITHOUT OVERLAPS). If false, all matching tuples are
- * compared against 'searchslot', which must contain a complete row.
+ * lossy GiST index for WITHOUT OVERLAPS) using the index's own equality
+ * semantics. If false, all matching tuples are compared against 'searchslot',
+ * which must contain a complete row.
  */
 bool
 RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
@@ -225,10 +226,7 @@ retry:
 		/*
 		 * Avoid expensive equality check if the index is primary key or
 		 * replica identity index, unless the index scan indicates that a
-		 * recheck is needed: a lossy index (e.g. a GiST index on a multirange
-		 * column of a WITHOUT OVERLAPS key) cannot prove an exact match, so
-		 * compare the key columns using the index's own equality semantics
-		 * (see repl_index_key_equal).
+		 * recheck is needed.
 		 */
 		if (!skipduplicates)
 		{
@@ -757,10 +755,7 @@ RelationFindDeletedTupleInfoByIndex(Relation rel, Oid idxoid,
 		/*
 		 * Avoid expensive equality check if the index is a primary key or
 		 * replica identity index, unless the index scan indicates that a
-		 * recheck is needed: a lossy index (e.g. a GiST index on a multirange
-		 * column of a WITHOUT OVERLAPS key) cannot prove an exact match, so
-		 * compare the key columns using the index's own equality semantics
-		 * (see repl_index_key_equal).
+		 * recheck is needed.
 		 */
 		if (!skipduplicates)
 		{
diff --git a/src/test/subscription/t/034_temporal.pl b/src/test/subscription/t/034_temporal.pl
index f8e662ef367..37b452e9297 100644
--- a/src/test/subscription/t/034_temporal.pl
+++ b/src/test/subscription/t/034_temporal.pl
@@ -635,7 +635,7 @@ drop_everything();
 # A temporal key over a multirange column is backed by a lossy GiST index: two
 # distinct multiranges that share a bounding range are indistinguishable in the
 # index, so an index probe can return candidate tuples that are not exact
-# matches.  The apply worker must recheck each candidate against the actual row
+# matches. The apply worker must recheck each candidate against the actual row
 # values, otherwise it could apply the change to the wrong row.
 # #################################
 
diff --git a/src/test/subscription/t/035_conflicts.pl b/src/test/subscription/t/035_conflicts.pl
index 03380ce2467..b2a7b29c116 100644
--- a/src/test/subscription/t/035_conflicts.pl
+++ b/src/test/subscription/t/035_conflicts.pl
@@ -447,9 +447,9 @@ $node_B->safe_psql('postgres', "DROP TABLE tab_defer");
 #
 # The publisher updates a row that never existed on the subscriber, while the
 # subscriber locally deleted a different row whose multirange key shares the
-# same bounding range.  Without a recheck, the dead "decoy" row would be
-# accepted as the update target and the conflict would be misreported as
-# update_deleted instead of update_missing.
+# same bounding range. Without a recheck, the deleted row would be accepted
+# as the update target and the conflict would be misreported as update_deleted
+# instead of update_missing.
 ###############################################################################
 
 # Publisher table holding the row that will be updated.
@@ -461,7 +461,7 @@ $node_B->safe_psql(
 	ALTER PUBLICATION tap_pub_B ADD TABLE tab_mr;
 ");
 
-# Subscriber table with the lossy replica identity index.  The index is
+# Subscriber table with the lossy replica identity index. The index is
 # created before the conflict detection slot's xmin is pinned below, so that
 # it can be used to find deleted tuples.
 $node_A->safe_psql('postgres',
@@ -475,8 +475,8 @@ $node_A->wait_for_subscription_sync($node_B, $subname_AB);
 
 # Pin the conflict detection slot's xmin: hold an open transaction with an
 # assigned transaction ID so that oldest_nonremovable_xid cannot advance past
-# it.  This keeps the decoy row deleted below retained and recently dead,
-# making the test independent of timing.
+# it. This keeps the row deleted below retained and recently dead, making the
+# test independent of timing.
 my $holder = $node_A->background_psql('postgres');
 $holder->query_safe('BEGIN');
 my $pinned_xid = $holder->query_safe('SELECT txid_current()');
@@ -487,7 +487,7 @@ ok( $node_A->poll_query_until(
 	),
 	"conflict detection slot xmin pinned at $pinned_xid");
 
-# Insert and locally delete the decoy row: a different multirange with the
+# Insert and locally delete the row: a different multirange with the
 # same bounding range as the publisher's key.
 $node_A->safe_psql(
 	'postgres', "
@@ -513,7 +513,7 @@ like(
 	'update target reported as missing');
 
 ok( $logfile !~ /conflict detected on relation "public.tab_mr": conflict=update_deleted/,
-	'decoy row sharing the bounding range is not reported as update_deleted');
+	'deleted row sharing the bounding range is not reported as update_deleted');
 
 # Clean up
 $holder->query_safe('COMMIT');