Kernel tests can't use path aliases on entities

This would be a great improvement and remove a massive WTF when testing this stuff! Changes overall look great and love the addition of alias test coverage <3

I did some thorough git archeology and the code block that this issue wants to remove from KernelTestBase::register() is there since the KernelTestBase class was added to core, #2336597: Convert path aliases to full featured entities just updated the comment inside that condition.

Also looked through the history of #3007661: Modernize the path alias system, and figured out that #3084983: Move all the code related to path aliases to a new (required) "path_alias" module and a couple of its followups are the issues that made the current MR possible. Additionally, the changes from this MR should help a lot with #3096092: Make "path_alias" module optional, so +1 on the smaller scope :)

All threads have been resolved and change makes sense.

Only concern is should there be a BC layer as this could break kernel tests in 10.2 right? But not sure since this is the test side.

I'm triaging RTBC issues. I read the IS and the comments here and in the MR. I didn't find any unanswered questions or other work to do.

Of special note is the git research done in #10. I noted that in the Issue Summary.

Leaving at RTBC.

Wow, I just stumbled upon this while working on #2418369: Internal URL handling (language prefixes, base://, ...). I couldn't understand what I was doing wrong. 😅 Happy to report that the patch from the MR got my kernel tests doing what I expected again. I'll try to closely review ASAP.

Thanks for the git archeology in #10, @amateescu, very helpful!

MR looks great. No complaints. All the changes look good and necessary. +1 to this change!

CR mostly looked great, although I added 1 line to the before example (for outbound path processing, too). No other complaints there.

Summary and title all look good.

I could imagine disruption if there are existing kernel tests in the wild (contrib / custom) that were relying on the broken behavior. But those are broken tests, so I think the disruption is a good idea. Hopefully the release managers agree. 😅

Thanks again!
-Derek

_{p.s. Crediting @mstrelan for the code, and the rest of us for reviews}
.

Committed to 11.x

Published change record.

Nice to see this gotcha fixed, I know it has bitten me before.

Sorry folks, I had to revert this - it caused fails in HEAD

https://git.drupalcode.org/project/drupal/-/pipelines/67196/test_report

Drupal\Tests\forum\Kernel\Migrate\d6\MigrateBlockTest::testBlockMigration
Drupal\Core\Database\DatabaseExceptionWrapper: SQLSTATE[42S02]: Base table or view not found: 1146 Table 'mysql.test98741461path_alias' doesn't exist: SELECT 1 AS "expression"
FROM
"test98741461path_alias" "base_table"
WHERE ("base_table"."status" = :db_condition_placeholder_0) AND ("base_table"."path" LIKE :db_condition_placeholder_1 ESCAPE '\\')
LIMIT 1 OFFSET 0; Array
(
    [:db_condition_placeholder_0] => 1
    [:db_condition_placeholder_1] => /admin%
)

Seems update to MigrateBlockTest is all green.

Compared the two patches locally and the diff was as expected

diff --git a/core/modules/forum/tests/src/Kernel/Migrate/d6/MigrateBlockTest.php b/core/modules/forum/tests/src/Kernel/Migrate/d6/MigrateBlockTest.php
index d018716c62c..aef056bdccc 100644
--- a/core/modules/forum/tests/src/Kernel/Migrate/d6/MigrateBlockTest.php
+++ b/core/modules/forum/tests/src/Kernel/Migrate/d6/MigrateBlockTest.php
@@ -30,6 +30,7 @@ class MigrateBlockTest extends MigrateDrupal6TestBase {
    */
   protected function setUp(): void {
     parent::setUp();
+    $this->installEntitySchema('path_alias');
 
     // Install the themes used for this test.
     $this->installEntitySchema('block_content');

Committed to 11.x, republished the change record