Fix regression in 2.7.1 when mysqli is used with discriminator column that is not a string

[mpdude]

In 2.7.1, #7974 was included as the fix for #7505. Basically, the change made was that when hydrating entities with inheritance, only the fields/data that belong to the particular inheritance type must be passed to UoW::createEntity().

To do so, the SimpleObjectHydrator will compare the discriminator column value against the list of known types/mappings.

orm/lib/Doctrine/ORM/Internal/Hydration/SimpleObjectHydrator.php

Lines 140 to 143 in 9273057

	// If we have inheritance in resultset, make sure the field belongs to the correct class
	if (isset($cacheKeyInfo['discriminatorValues']) && ! in_array($discrColumnValue, $cacheKeyInfo['discriminatorValues'], true)) {
	continue;
	}

While the descriminator types in $cacheKeyInfo['discriminatorValues'] are always strings, the actual $discrColumnValue may be an int, at least when the discriminator column is defined as int and the mysqli driver is used.

The problem seems not to occur with pdo_mysql: In that case, the values returned from the result set/query are (always?) strings.

My suggested fix is to (string) cast the discriminator value, as it already happens in the AbstractHydrator:

orm/lib/Doctrine/ORM/Internal/Hydration/AbstractHydrator.php

Lines 300 to 306 in 9273057

	// If there are field name collisions in the child class, then we need
	// to only hydrate if we are looking at the correct discriminator value
	if (isset($cacheKeyInfo['discriminatorColumn'], $data[$cacheKeyInfo['discriminatorColumn']])
	&& ! in_array((string) $data[$cacheKeyInfo['discriminatorColumn']], $cacheKeyInfo['discriminatorValues'], true)
	) {
	break;
	}

Update:

It seems Travis is running MySQL tests with pdo_mysql only, so the test case does not even show the issue. I am also sure that the issue would have surfaced with the existing tests.

I still believe the fix is correct, given the similarity to the code in AbstractHydrator.

Anyway, could someone advise if this pdo_mysql/mysqli discrepancy is something worth looking into?

[mpdude]

What about running tests for mysqli?

I understand that the test matrix is already very complex and resource intensive.

But unless we run mysqli, the regression test is useless and we’re not going to find such bugs.