Document that select_for_update() doesn't lock tables when their columns are not selected.

[Hannes Ljungberg]

The existing test "test_for_update_of_self_when_self_is_not_selected" can be extended to demonstrate this behaviour:

diff --git a/tests/select_for_update/tests.py b/tests/select_for_update/tests.py
index 705975b760..00fa102d33 100644
--- a/tests/select_for_update/tests.py
+++ b/tests/select_for_update/tests.py
@@ -240,9 +240,15 @@ class SelectForUpdateTests(TransactionTestCase):
         select_for_update(of=['self']) when the only columns selected are from
         related tables.
         """
-        with transaction.atomic():
+        with transaction.atomic(), CaptureQueriesContext(connection) as ctx:
             values = list(Person.objects.select_related('born').select_for_update(of=('self',)).values('born__name'))
         self.assertEqual(values, [{'born__name': self.city1.name}])
+        if connection.features.select_for_update_of_column:
+            expected = ['select_for_update_person"."id']
+        else:
+            expected = ['select_for_update_person']
+        expected = [connection.ops.quote_name(value) for value in expected]
+        self.assertTrue(self.has_for_update_sql(ctx.captured_queries, of=expected))

     @skipUnlessDBFeature('has_select_for_update_nowait')
     def test_nowait_raises_error_on_block(self):

[Hannes Ljungberg]

Tried to understand what's happening in SQLCompiler.get_select_for_update_of_arguments and it appears like it's dependant on what tables and columns have been SELECT:ed. Looks like we only need to do this because Oracle want columns instead of tables. I wonder if we could base this on the JOIN:ed tables instead and grab the column from the predicate for Oracle?

[Mariusz Felisiak]

IMO this is an expected behavior, we changed it in the ​PR.

[Hannes Ljungberg]

Thanks, missed that PR. IMO it's a bit unexpected that the selected fields might end up affecting FOR UPDATE. What do you think about adding a small note about this in the docs?

[Mariusz Felisiak]

Replying to Hannes Ljungberg:

Thanks, missed that PR. IMO it's a bit unexpected that the selected fields might end up affecting FOR UPDATE. What do you think about adding a small note about this in the docs?

Sure, feel-free to propose a clarification.

[Jacob Walls]

​PR

[GitHub <noreply@…>]

In d400b08a:

Fixed #32888 -- Doc'd that select_for_update() only locks tables with selected columns.

[Mariusz Felisiak <felisiak.mariusz@…>]

In 816e8093:

[4.0.x] Fixed #32888 -- Doc'd that select_for_update() only locks tables with selected columns.

Backport of d400b08a8baa697905daadd47a6ba12336e93336 from main