PostgreSQL Duplicate table error for alter index queries

Created a fix for the bug including added tests.

To do that I also had to add an API-addition for the method Drupal\Core\Database\Schema::tableExists(). I added a second parameter to the method $add_prefix which defaults to TRUE. Also added testing for that.

Fixing the failure for MySQL. Therefore added the internal method Drupal\Core\Database\Connection::getQuoteIdentifiers().

Updated the IS.

Added a CR.

Updated the table names in the added test.

I have read the patch and this looks fine to me and the test shows it fixes the problem in the IS.

The patch still applies to 9.2.x.

My only comment is about readability. After applying the patch and reading the change in \Drupal\Core\Database\Driver\mysql\Schema::tableExists I found myself constantly checking the two queryRange statements to see the difference. So, I was thinking it help to modify that a bit.

+++ b/core/lib/Drupal/Core/Database/Driver/mysql/Schema.php
@@ -680,7 +680,14 @@ public function tableExists($table) {
+        $this->connection->queryRange("SELECT 1 FROM {" . $table . "}", 0, 1);
...
+        $this->connection->queryRange('SELECT 1 FROM ' . $table, 0, 1);

Maybe modify this so that only the queryRange statement is inside the try. Something like $this->connection->queryRange("SELECT 1 FROM $name", 0, 1);

Converted to an MR and hide patches.

There is a failing test, Drupal\Tests\migrate_drupal\Unit\MigrationConfigurationTraitTest.

Small nitpicky stuff.

@smustgrave, thanks. I should have done a self review and picked those up before setting to needs review.

I removed the method getQuoteIdentifiers which was in the patch but no longer used in the MR because there is no longer a tableExists method specific for mysql.

Known random failure on

    Drupal\Tests\settings_tray\FunctionalJavascript\SettingsTrayBlockFormTest::testBlocks
    WebDriver\Exception\CurlExec: Curl error thrown for http POST to

Therefore setting back to Needs review.

Minor remark on the MR.

Could we run the pipeline for PostgreSQL for this MR?

@daffie, thanks for the review.

I made the change requested and ran against MySQL, PostgreSQL and SQLite.

Unrelated failure in Drupal\Tests\field_ui\FunctionalJavascript\ManageFieldsTest

Appears all feedback has been addressed.

Simplify title a bit.

Re-titling again since this isn't really about updating from 8.5

Committed to 11.x

Didn't backport to 10.2.x due to the risk of disruption from the signature change

Published change record

Thanks all

This is unfortunately a breaking change for Drupal 10.3, isn't it?

Re: #31, no BC issue because the new parameter has a default value which preserves current behavior.

@bradjones1 for those calling the method, that's OK. But there is e.g. the sqlsrv module which extends that method, see \Drupal\sqlsrv\Driver\Database\sqlsrv\Schema::tableExists and suddenly becomes incompatible.

As the abstract class \Drupal\Core\Database\Schema isn't marked internal, it feels as if extending it is what's supposed to be done.