7. Working with Objects

In this chapter we will help you understand the EntityManager and the UnitOfWork. A Unit of Work is similar to an object-level transaction. A new Unit of Work is implicitly started when an EntityManager is initially created or after EntityManager#flush() has been invoked. A Unit of Work is committed (and a new one started) by invoking EntityManager#flush().

A Unit of Work can be manually closed by calling EntityManager#close(). Any changes to objects within this Unit of Work that have not yet been persisted are lost.


It is very important to understand that only EntityManager#flush() ever causes write operations against the database to be executed. Any other methods such as EntityManager#persist($entity) or EntityManager#remove($entity) only notify the UnitOfWork to perform these operations during flush.

Not calling EntityManager#flush() will lead to all changes during that request being lost.

7.1. Entities and the Identity Map

Entities are objects with identity. Their identity has a conceptual meaning inside your domain. In a CMS application each article has a unique id. You can uniquely identify each article by that id.

Take the following example, where you find an article with the headline “Hello World” with the ID 1234:

$article = $entityManager->find('CMS\Article', 1234);
$article->setHeadline('Hello World dude!');

$article2 = $entityManager->find('CMS\Article', 1234);
echo $article2->getHeadline();

In this case the Article is accessed from the entity manager twice, but modified in between. Doctrine 2 realizes this and will only ever give you access to one instance of the Article with ID 1234, no matter how often do you retrieve it from the EntityManager and even no matter what kind of Query method you are using (find, Repository Finder or DQL). This is called “Identity Map” pattern, which means Doctrine keeps a map of each entity and ids that have been retrieved per PHP request and keeps returning you the same instances.

In the previous example the echo prints “Hello World dude!” to the screen. You can even verify that $article and $article2 are indeed pointing to the same instance by running the following code:

if ($article === $article2) {
    echo "Yes we are the same!";

Sometimes you want to clear the identity map of an EntityManager to start over. We use this regularly in our unit-tests to enforce loading objects from the database again instead of serving them from the identity map. You can call EntityManager#clear() to achieve this result.

7.2. Entity Object Graph Traversal

Although Doctrine allows for a complete separation of your domain model (Entity classes) there will never be a situation where objects are “missing” when traversing associations. You can walk all the associations inside your entity models as deep as you want.

Take the following example of a single Article entity fetched from newly opened EntityManager.

/** @Entity */
class Article
    /** @Id @Column(type="integer") @GeneratedValue */
    private $id;

    /** @Column(type="string") */
    private $headline;

    /** @ManyToOne(targetEntity="User") */
    private $author;

    /** @OneToMany(targetEntity="Comment", mappedBy="article") */
    private $comments;

    public function __construct()
        $this->comments = new ArrayCollection();

    public function getAuthor() { return $this->author; }
    public function getComments() { return $this->comments; }

$article = $em->find('Article', 1);

This code only retrieves the Article instance with id 1 executing a single SELECT statement against the articles table in the database. You can still access the associated properties author and comments and the associated objects they contain.

This works by utilizing the lazy loading pattern. Instead of passing you back a real Author instance and a collection of comments Doctrine will create proxy instances for you. Only if you access these proxies for the first time they will go through the EntityManager and load their state from the database.

This lazy-loading process happens behind the scenes, hidden from your code. See the following code:

$article = $em->find('Article', 1);

// accessing a method of the user instance triggers the lazy-load
echo "Author: " . $article->getAuthor()->getName() . "\n";

// Lazy Loading Proxies pass instanceof tests:
if ($article->getAuthor() instanceof User) {
    // a User Proxy is a generated "UserProxy" class

// accessing the comments as an iterator triggers the lazy-load
// retrieving ALL the comments of this article from the database
// using a single SELECT statement
foreach ($article->getComments() as $comment) {
    echo $comment->getText() . "\n\n";

// Article::$comments passes instanceof tests for the Collection interface
// But it will NOT pass for the ArrayCollection interface
if ($article->getComments() instanceof \Doctrine\Common\Collections\Collection) {
    echo "This will always be true!";

A slice of the generated proxy classes code looks like the following piece of code. A real proxy class override all non-identity non-transient object state at instantiation time in order to enable lazy-loading mechanisms:

class UserProxyHASH extends User implements GhostObjectInterface
    // ... generated code

    public static function staticProxyConstructor($initializer)
        // ... generated code

    private function callInitializerHASH($methodName, array $parameters)
        // ... generated code

    // ... generated code


Traversing the object graph for parts that are lazy-loaded will easily trigger lots of SQL queries and will perform badly if used to heavily. Make sure to use DQL to fetch-join all the parts of the object-graph that you need as efficiently as possible.

7.3. Persisting entities

An entity can be made persistent by passing it to the EntityManager#persist($entity) method. By applying the persist operation on some entity, that entity becomes MANAGED, which means that its persistence is from now on managed by an EntityManager. As a result the persistent state of such an entity will subsequently be properly synchronized with the database when EntityManager#flush() is invoked.


Invoking the persist method on an entity does NOT cause an immediate SQL INSERT to be issued on the database. Doctrine applies a strategy called “transactional write-behind”, which means that it will delay most SQL commands until EntityManager#flush() is invoked which will then issue all necessary SQL statements to synchronize your objects with the database in the most efficient way and a single, short transaction, taking care of maintaining referential integrity.


$user = new User;


Generated entity identifiers / primary keys are guaranteed to be available after the next successful flush operation that involves the entity in question. You can not rely on a generated identifier to be available directly after invoking persist. The inverse is also true. You can not rely on a generated identifier being not available after a failed flush operation.

The semantics of the persist operation, applied on an entity X, are as follows:

  • If X is a new entity, it becomes managed. The entity X will be entered into the database as a result of the flush operation.
  • If X is a preexisting managed entity, it is ignored by the persist operation. However, the persist operation is cascaded to entities referenced by X, if the relationships from X to these other entities are mapped with cascade=PERSIST or cascade=ALL (see “Transitive Persistence”).
  • If X is a removed entity, it becomes managed.
  • If X is a detached entity, an exception will be thrown on flush.

7.4. Removing entities

An entity can be removed from persistent storage by passing it to the EntityManager#remove($entity) method. By applying the remove operation on some entity, that entity becomes REMOVED, which means that its persistent state will be deleted once EntityManager#flush() is invoked.


Just like persist, invoking remove on an entity does NOT cause an immediate SQL DELETE to be issued on the database. The entity will be deleted on the next invocation of EntityManager#flush() that involves that entity. This means that entities scheduled for removal can still be queried for and appear in query and collection results. See the section on Database and UnitOfWork Out-Of-Sync for more information.



The semantics of the remove operation, applied to an entity X are as follows:

  • If X is a new entity, it is ignored by the remove operation. However, the remove operation is cascaded to entities referenced by X, if the relationship from X to these other entities is mapped with cascade=REMOVE or cascade=ALL (see “Transitive Persistence”).
  • If X is a managed entity, the remove operation causes it to become removed. The remove operation is cascaded to entities referenced by X, if the relationships from X to these other entities is mapped with cascade=REMOVE or cascade=ALL (see “Transitive Persistence”).
  • If X is a detached entity, an InvalidArgumentException will be thrown.
  • If X is a removed entity, it is ignored by the remove operation.
  • A removed entity X will be removed from the database as a result of the flush operation.

After an entity has been removed its in-memory state is the same as before the removal, except for generated identifiers.

Removing an entity will also automatically delete any existing records in many-to-many join tables that link this entity. The action taken depends on the value of the @joinColumn mapping attribute “onDelete”. Either Doctrine issues a dedicated DELETE statement for records of each join table or it depends on the foreign key semantics of onDelete=”CASCADE”.

Deleting an object with all its associated objects can be achieved in multiple ways with very different performance impacts.

  1. If an association is marked as CASCADE=REMOVE Doctrine 2 will fetch this association. If its a Single association it will pass this entity to ´EntityManager#remove()``. If the association is a collection, Doctrine will loop over all its elements and pass them to``EntityManager#remove()`. In both cases the cascade remove semantics are applied recursively. For large object graphs this removal strategy can be very costly.
  2. Using a DQL DELETE statement allows you to delete multiple entities of a type with a single command and without hydrating these entities. This can be very efficient to delete large object graphs from the database.
  3. Using foreign key semantics onDelete="CASCADE" can force the database to remove all associated objects internally. This strategy is a bit tricky to get right but can be very powerful and fast. You should be aware however that using strategy 1 (CASCADE=REMOVE) completely by-passes any foreign key onDelete=CASCADE option, because Doctrine will fetch and remove all associated entities explicitly nevertheless.

7.5. Detaching entities

All entities are detached from an EntityManager and thus no longer managed by it after invoking the EntityManager#clear() method. Changes made to the detached entities, if any (including their removal), will not be synchronized to the database after they have been detached.

Doctrine will not hold on to any references to detached entities.



The semantics of the detach operation, applied to an entity X are as follows:

  • If X is a managed entity, the clear operation causes it to become detached. Entities which previously referenced X will continue to reference X.
  • If X is a new or detached entity, it is ignored by the detach operation.
  • If X is a removed entity, it will become detached, and therefore no longer scheduled to be removed. Entities which previously referenced X will continue to reference X.

There are several situations in which an entity is detached automatically:

  • When serializing an entity. The entity retrieved upon subsequent unserialization will be detached (This is the case for all entities that are serialized and stored in some cache).

7.6. Synchronization with the Database

The state of persistent entities is synchronized with the database on flush of an EntityManager which commits the underlying UnitOfWork. The synchronization involves writing any updates to persistent entities and their relationships to the database. Thereby bidirectional relationships are persisted based on the references held by the owning side of the relationship as explained in the Association Mapping chapter.

When EntityManager#flush() is called, Doctrine inspects all managed, new and removed entities and will perform the following operations.

7.6.1. Effects of Database and UnitOfWork being Out-Of-Sync

As soon as you begin to change the state of entities, call persist or remove the contents of the UnitOfWork and the database will drive out of sync. They can only be synchronized by calling EntityManager#flush(). This section describes the effects of database and UnitOfWork being out of sync.

  • Entities that are scheduled for removal can still be queried from the database. They are returned from DQL and Repository queries and are visible in collections.
  • Entities that are passed to EntityManager#persist do not turn up in query results.
  • Entities that have changed will not be overwritten with the state from the database. This is because the identity map will detect the construction of an already existing entity and assumes its the most up to date version.

EntityManager#flush() is never called implicitly by Doctrine. You always have to trigger it manually.

7.6.2. Synchronizing New and Managed Entities

The flush operation applies to a managed entity with the following semantics:

  • The entity itself is synchronized to the database using a SQL UPDATE statement, only if at least one persistent field has changed.
  • No SQL updates are executed if the entity did not change.

The flush operation applies to a new entity with the following semantics:

  • The entity itself is synchronized to the database using a SQL INSERT statement.

For all (initialized) relationships of the new or managed entity the following semantics apply to each associated entity X:

  • If X is new and persist operations are configured to cascade on the relationship, X will be persisted.
  • If X is new and no persist operations are configured to cascade on the relationship, an exception will be thrown as this indicates a programming error.
  • If X is removed and persist operations are configured to cascade on the relationship, an exception will be thrown as this indicates a programming error (X would be re-persisted by the cascade).
  • If X is detached and persist operations are configured to cascade on the relationship, an exception will be thrown (This is semantically the same as passing X to persist()).

7.6.3. Synchronizing Removed Entities

The flush operation applies to a removed entity by deleting its persistent state from the database. No cascade options are relevant for removed entities on flush, the cascade remove option is already executed during EntityManager#remove($entity).

7.6.4. The size of a Unit of Work

The size of a Unit of Work mainly refers to the number of managed entities at a particular point in time.

7.6.5. The cost of flushing

How costly a flush operation is, mainly depends on two factors:

  • The size of the EntityManager’s current UnitOfWork.
  • The configured change tracking policies

You can get the size of a UnitOfWork as follows:

$uowSize = $em->getUnitOfWork()->size();

The size represents the number of managed entities in the Unit of Work. This size affects the performance of flush() operations due to change tracking (see “Change Tracking Policies”) and, of course, memory consumption, so you may want to check it from time to time during development.


Do not invoke flush after every change to an entity or every single invocation of persist/remove/refresh/… This is an anti-pattern and unnecessarily reduces the performance of your application. Instead, form units of work that operate on your objects and call flush when you are done. While serving a single HTTP request there should be usually no need for invoking flush more than 0-2 times.

7.6.6. Direct access to a Unit of Work

You can get direct access to the Unit of Work by calling EntityManager#getUnitOfWork(). This will return the UnitOfWork instance the EntityManager is currently using.

$uow = $em->getUnitOfWork();


Directly manipulating a UnitOfWork is not recommended. When working directly with the UnitOfWork API, respect methods marked as INTERNAL by not using them and carefully read the API documentation.

7.6.7. Entity State

As outlined in the architecture overview an entity can be in one of four possible states: NEW, MANAGED, REMOVED, DETACHED. If you explicitly need to find out what the current state of an entity is in the context of a certain EntityManager you can ask the underlying UnitOfWork:

switch ($em->getUnitOfWork()->getEntityState($entity)) {
    case UnitOfWork::STATE_MANAGED:
    case UnitOfWork::STATE_REMOVED:
    case UnitOfWork::STATE_DETACHED:
    case UnitOfWork::STATE_NEW:

An entity is in MANAGED state if it is associated with an EntityManager and it is not REMOVED.

An entity is in REMOVED state after it has been passed to EntityManager#remove() until the next flush operation of the same EntityManager. A REMOVED entity is still associated with an EntityManager until the next flush operation.

An entity is in DETACHED state if it has persistent state and identity but is currently not associated with an EntityManager.

An entity is in NEW state if has no persistent state and identity and is not associated with an EntityManager (for example those just created via the “new” operator).

7.7. Querying

Doctrine 2 provides the following ways, in increasing level of power and flexibility, to query for persistent objects. You should always start with the simplest one that suits your needs.

7.7.1. By Primary Key

The most basic way to query for a persistent object is by its identifier / primary key using the EntityManager#find($entityName, $id) method. Here is an example:

// $em instanceof EntityManager
$user = $em->find('MyProject\Domain\User', $id);

The return value is either the found entity instance or null if no instance could be found with the given identifier.

Essentially, EntityManager#find() is just a shortcut for the following:

// $em instanceof EntityManager
$user = $em->getRepository('MyProject\Domain\User')->find($id);

EntityManager#getRepository($entityName) returns a repository object which provides many ways to retrieve entities of the specified type. By default, the repository instance is of type Doctrine\ORM\EntityRepository. You can also use custom repository classes as shown later.

7.7.2. By Simple Conditions

To query for one or more entities based on several conditions that form a logical conjunction, use the findBy and findOneBy methods on a repository as follows:

// $em instanceof EntityManager

// All users that are 20 years old
$users = $em->getRepository('MyProject\Domain\User')->findBy(array('age' => 20));

// All users that are 20 years old and have a surname of 'Miller'
$users = $em->getRepository('MyProject\Domain\User')->findBy(array('age' => 20, 'surname' => 'Miller'));

// A single user by its nickname
$user = $em->getRepository('MyProject\Domain\User')->findOneBy(array('nickname' => 'romanb'));

You can also load by owning side associations through the repository:

$number = $em->find('MyProject\Domain\Phonenumber', 1234);
$user = $em->getRepository('MyProject\Domain\User')->findOneBy(array('phone' => $number->getId()));

The EntityRepository#findBy() method additionally accepts orderings, limit and offset as second to fourth parameters:

$tenUsers = $em->getRepository('MyProject\Domain\User')->findBy(array('age' => 20), array('name' => 'ASC'), 10, 0);

If you pass an array of values Doctrine will convert the query into a WHERE field IN (..) query automatically:

$users = $em->getRepository('MyProject\Domain\User')->findBy(array('age' => array(20, 30, 40)));
// translates roughly to: SELECT * FROM users WHERE age IN (20, 30, 40)

An EntityRepository also provides a mechanism for more concise calls through its use of __call. Thus, the following two examples are equivalent:

// A single user by its nickname
$user = $em->getRepository('MyProject\Domain\User')->findOneBy(array('nickname' => 'romanb'));

// A single user by its nickname (__call magic)
$user = $em->getRepository('MyProject\Domain\User')->findOneByNickname('romanb');

Additionally, you can just count the result of the provided conditions when you don’t really need the data:

// Check there is no user with nickname
$availableNickname = 0 === $em->getRepository('MyProject\Domain\User')->count(['nickname' => 'nonexistent']);

7.7.3. By Criteria

New in version 2.3.

The Repository implement the Doctrine\Common\Collections\Selectable interface. That means you can build Doctrine\Common\Collections\Criteria and pass them to the matching($criteria) method.

See section Filtering collections of chapter Working with Associations

7.7.4. By Eager Loading

Whenever you query for an entity that has persistent associations and these associations are mapped as EAGER, they will automatically be loaded together with the entity being queried and is thus immediately available to your application.

7.7.5. By Lazy Loading

Whenever you have a managed entity instance at hand, you can traverse and use any associations of that entity that are configured LAZY as if they were in-memory already. Doctrine will automatically load the associated objects on demand through the concept of lazy-loading.

7.7.6. By DQL

The most powerful and flexible method to query for persistent objects is the Doctrine Query Language, an object query language. DQL enables you to query for persistent objects in the language of objects. DQL understands classes, fields, inheritance and associations. DQL is syntactically very similar to the familiar SQL but it is not SQL.

A DQL query is represented by an instance of the Doctrine\ORM\Query class. You create a query using EntityManager#createQuery($dql). Here is a simple example:

// $em instanceof EntityManager

// All users with an age between 20 and 30 (inclusive).
$q = $em->createQuery("select u from MyDomain\Model\User u where u.age >= 20 and u.age <= 30");
$users = $q->getResult();

Note that this query contains no knowledge about the relational schema, only about the object model. DQL supports positional as well as named parameters, many functions, (fetch) joins, aggregates, subqueries and much more. Detailed information about DQL and its syntax as well as the Doctrine class can be found in the dedicated chapter. For programmatically building up queries based on conditions that are only known at runtime, Doctrine provides the special Doctrine\ORM\QueryBuilder class. More information on constructing queries with a QueryBuilder can be found in Query Builder chapter.

7.7.7. By Native Queries

As an alternative to DQL or as a fallback for special SQL statements native queries can be used. Native queries are built by using a hand-crafted SQL query and a ResultSetMapping that describes how the SQL result set should be transformed by Doctrine. More information about native queries can be found in the dedicated chapter.

7.7.8. Custom Repositories

By default the EntityManager returns a default implementation of Doctrine\ORM\EntityRepository when you call EntityManager#getRepository($entityClass). You can overwrite this behaviour by specifying the class name of your own Entity Repository in the Annotation or XML metadata. In large applications that require lots of specialized DQL queries using a custom repository is one recommended way of grouping these queries in a central location.

namespace MyDomain\Model;

use Doctrine\ORM\EntityRepository;
use Doctrine\ORM\Mapping as ORM;

 * @ORM\Entity(repositoryClass="MyDomain\Model\UserRepository")
class User


class UserRepository extends EntityRepository
    public function getAllAdminUsers()
        return $this->em->createQuery('SELECT u FROM MyDomain\Model\User u WHERE u.status = "admin"')

You can access your repository now by calling:

// $em instanceof EntityManager

$admins = $em->getRepository('MyDomain\Model\User')->getAllAdminUsers();
Fork me on GitHub