If you use in your MySQL database default values, indexes or string primary key, you get incorrect mapping by mapping generator. For get it - just use in database one or more from listed abilities, generate mapping for that and then try to dump-sql with schema-tool:update.

Hope you fix it. Tnx!

I have entity A and B, the relationship between A and B is many-to-many. which means Doctrine2 will generate a middle table called AB for me.

entity A:

class Station {
    /**
     * @ORM\ManyToMany(targetEntity="Fun", mappedBy="stations")
     */
    protected $funs;
}

entity B:

class Fun {
    /**
     * @ORM\ManyToMany(targetEntity="Station", inversedBy="funs")
     * @ORM\JoinTable(name="stations_have_funs")
     */
    protected $stations;
}

the schema of middle table stations_have_funs:

CREATE TABLE `stations_have_funs` (
  `fun_id` int(11) NOT NULL,
  `station_id` int(11) NOT NULL,
  PRIMARY KEY (`fun_id`,`station_id`),
  KEY `IDX_45C921911CA4BE49` (`fun_id`),
  KEY `IDX_45C9219121BDB235` (`station_id`),
  CONSTRAINT `FK_45C921911CA4BE49` FOREIGN KEY (`fun_id`) REFERENCES `funs` (`id`) ON DELETE CASCADE,
  CONSTRAINT `FK_45C9219121BDB235` FOREIGN KEY (`station_id`) REFERENCES `stations` (`id`) ON DELETE CASCADE
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;

I noticed that there are 2 useless index(fun_id and station_id). Since fun_id and station_id are the primary key of this table. Do we really need 2 extra/duplicated index ?

This issue is created automatically through a Github pull request on behalf of TorbenBr:

Url: https://github.com/doctrine/doctrine2/pull/674

Message:

In order to simplify and speed up the sql generation
an ANSI quote strategy would be useful.

The implementation would be something like :

<?php
class AnsiQuoteStrategy implements \Doctrine\ORM\Mapping\QuoteStrategy
{
    /**
     * {@inheritdoc}
     */
    public function getColumnName($fieldName, ClassMetadata $class, AbstractPlatform $platform)
    {
        return $class->fieldMappings[$fieldName]['columnName'];
    }

    /**
     * {@inheritdoc}
     */
    public function getTableName(ClassMetadata $class, AbstractPlatform $platform)
    {
        return $class->table['name'];
    }

    /**
     * {@inheritdoc}
     */
    public function getSequenceName(array $definition, ClassMetadata $class, AbstractPlatform $platform)
    {
        return $definition['sequenceName'];
    }

    /**
     * {@inheritdoc}
     */
    public function getJoinColumnName(array $joinColumn, ClassMetadata $class, AbstractPlatform $platform)
    {
        return $joinColumn['name'];
    }

    /**
     * {@inheritdoc}
     */
    public function getReferencedJoinColumnName(array $joinColumn, ClassMetadata $class, AbstractPlatform $platform)
    {
        return $joinColumn['referencedColumnName'];
    }

    /**
     * {@inheritdoc}
     */
    public function getJoinTableName(array $association, ClassMetadata $class, AbstractPlatform $platform)
    {
        return $association['joinTable']['name'];
    }

    /**
     * {@inheritdoc}
     */
    public function getIdentifierColumnNames(ClassMetadata $class, AbstractPlatform $platform)
    {
        return $class->identifier;
    }

    /**
     * {@inheritdoc}
     */
    public function getColumnAlias($columnName, $counter, AbstractPlatform $platform, ClassMetadata $class = null)
    {
        return $platform->getSQLResultCashing($columnName . $counter);
    }
}

class ProductCategory
{
    /**
     * Store
     *
     * @var Store
     *
     * @ORM\ManyToOne(targetEntity="Store")
     * @ORM\JoinColumn(name="store_id", referencedColumnName="store_id")
     */
    private $Store;

    /**
     * storeId (for ordering in Product::ProductCategories only)
     *
     * @var integer
     *
     * @ORM\Column(name="store_id", type="integer")
     */
    private $storeId;
...

class Product
{
    /**
     * Associated categories
     *
     * @var \Doctrine\Common\Collections\Collection
     *
     * @ORM\OneToMany(targetEntity="ProductCategory", mappedBy="Product")
     * @ORM\OrderBy({"storeId"="ASC"})
     */
    private $ProductCategories;
}
}

If it is possible now to sort the ProductCategories collection by the storeId field, it should also be possible to sort them by the Store association. Currently a set of two fields is required: Store as a regular Many-To-One association and if a need arises to be able to use it to sort the One-To-Many collections then storeId needs to be added to the ProductCategory entity. In that case the ProductCategory entity does not pass the schema validation but is perfectly usable.

This should be allowed:

class Product
{
    /**
     * Associated categories
     *
     * @var \Doctrine\Common\Collections\Collection
     *
     * @ORM\OneToMany(targetEntity="ProductCategory", mappedBy="Product")
     * @ORM\OrderBy({"Store"="ASC"})
     */
    private $ProductCategories;
}

Hi.
In mysql db table i have > 200,000 items.
I use native doctrine pagination for paging the items list.
But generated query that gets ids for items list in paging works more then 150 sec on my workstation

SELECT DISTINCT id0 FROM (SELECT m0_.id AS id0, m0_.title AS title1, m0_.text AS text2, m0_.price AS price3, m0_.originalPrice AS originalPrice4, m0_.condition_type AS condition_type5, m0_.image_1 AS image_16, m0_.image_2 AS image_27, m0_.image_3 AS image_38, m0_.image_4 AS image_49, m0_.image_5 AS image_510, m0_.video AS video11, m0_.contact_email AS contact_email12, m0_.contact_name AS contact_name13, m0_.contact_phone AS contact_phone14, m0_.contact_type AS contact_type15, m0_.published AS published16, m0_.type AS type17, m0_.status AS status18, m0_.highlight AS highlight19, m0_.urgent AS urgent20, m0_.topads AS topads21, m0_.period AS period22, m0_.hits AS hits23, m0_.ip AS ip24, m0_.created_at AS created_at25, m0_.updated_at AS updated_at26 FROM milla_message m0_ INNER JOIN milla_currency m1_ ON m0_.currency_id = m1_.id INNER JOIN milla_category m2_ ON m0_.category_id = m2_.id INNER JOIN milla_region m3_ ON m0_.region_id = m3_.id INNER JOIN milla_city m4_ ON m0_.city_id = m4_.id WHERE m0_.status = 1 ORDER BY m0_.published DESC) dctrn_result LIMIT 20 OFFSET 0

source code
https://github.com/doctrine/doctrine2/blob/master/lib/Doctrine/ORM/Tools/Pagination/LimitSubqueryOutputWalker.php#L141

why SELECT DISTINCT %s FROM (%s) dctrn_result ???
why not SELECT DISTINCT m0_.id AS id0 FROM milla_message m0_ WHERE m0_.status = 1 ORDER BY m0_.published DESC LIMIT 20 OFFSET 0

I don't want you to report something that isn't a bug.
But I read about problems with doctrine2 and mssql uniqueid's.
So First I asked a question at stackoverflow. No one could help me, and the only one who gave me a comment thought the same then me, that it looks like a bug.

If it isn't a bug I'm very sorry for this issue report.

Issue as reported in
http://stackoverflow.com/questions/15368082/refresh-uniqueidentifier-id-from-mssql-of-inserted-entity-in-doctrine2

I've seen that when you create an entity with a composite foreign key as identifier it cannot be flushed until the related entities are already flushed to the database and not just persisted.

It would be nice to let the user flush all the entities together and just INSERT first the related entities to get the ID and then use that to INSERT the entity with composite foreign keys.

I'm going to create a pull request with the failing test.

When executing a simple query with field names in SELECT clause, it is not possible to map field to numeric indexes.

This is an example that i would imagine to be useful:

SELECT c.id AS 0, c.name AS 1, l.text AS 2 FROM Category c LEFT JOIN c.label l

Thus, the resulting results could be numeric indexed array. It is useful for many situations: when working with an API which expects such arrays, or when using list to assign result fields to variables directly.

Query::HYDRATE_SCALAR does not achieve this, as one could think at first glance.

Unfortunately DatabaseDriver::getClassNameForTable() is declared as private method, which makes it quite difficult to change the naming strategy for reverse engeneering.

IMO this sould be declared protected. An even better way would be to extend the interface of the Naming Strategy objects to support the reverse direction:

classToTableName -> tableToClassName
propertyToColumnName -> columnToPropertyName.

This way we would have a consistent name-mapping

ResultSetMappingBuilder does not support inherited fields. For example, calling ResultSetMappingBuilder::addRootEntityFromClassMetadata($class, $alias) throws an exception to say this.

I was wondering if there were any reasons as to why this would be difficult to implement? I haven't had an extensive look at Doctrine's source but it feels like this has been not implemented on purpose.

Thanks

In Symfony2, when updating the collation option of a table, the schema tool doesn't recognize the change:

Changing from:

 
* @ORM\Table()

To:

 
* @ORM\Table(options={"collate"="utf8_swedish_ci"})

Results in:

 
$ php app/console doctrine:schema:update --dump-sql
Nothing to update - your database is already in sync with the current entity metadata.

It would be better if doctrine generated "isEnabled()" instead of "getEnabled()" for boolean field in entity classes. Because, it is more meaningful.

The documentation at http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/faq.html#how-do-i-set-the-charset-and-collation-for-mysql-tables clearly states that the collation should be set at database level and then inherited for all tables created.

Digging through the code and reading this issue http://www.doctrine-project.org/jira/browse/DDC-2139 it's clear that this is no longer the case.

entity A have many entity B

If DQL is something like "select A, B from A join B order by A.field1, B.field2"
Then Paginator retrieve different Id by a query like :
"select distinct a.id, a.field1, b.field2 from ( ....) order by a.field1 ASC , b.field2 ASC limit 15 offset 0"

so, if a entity A have 20 entity B (and these sub-entity have all a different b.field2) then there's only 1 A retrieved

In using a database-first approach utilizing orm:convert-mapping to generate xml, the validation and schema-tool reports that my composite primary key (ex. Columns A, C, B) be dropped and added in the order in which the mapping appears in the xml (ex. Columns A, B, C).

These columns are not auto-increment and are simply a mixture of int and varchar.

When executing partial DQL queries it may be important to keep given order of columns e.g. for "pairs" hydrator when first column of a pair is used as a key and second - as value. For example query like this:

select partial u.{id,name} from my:User u

will expect "id" to be first in resulted set and "name" to be second and not vice versa.

However Doctrine parses this part of statement via iterating over fields mapping from entity's class metadata (as can be seen in Doctrine\ORM\Query\SqlWalker::walkSelectExpression()):

foreach ($class->fieldMappings as $fieldName => $mapping) {
  if ($partialFieldSet && ! in_array($fieldName, $partialFieldSet)) {
    continue;
  }
  ...

and hence given columns order preserving is not guaranteed.

https://github.com/whiteoctober/Pagerfanta/issues/69

<?php
$query = $em->getRepository('M\E\Q')
->createQueryBuilder('q')
->select('q', 'SUM(q.price) AS amount')
->where('q.id IN(19, 20, 22)')
->groupBy('q.customer')
;

$pager = new Pagerfanta(new DoctrineORMAdapter($query));
$pager->setMaxPerPage(30);
$pager->setCurrentPage($request->query->get('page', 1));

$result = $pager->getCurrentPageResults();
print_r($result[0]['amount']); // 156.71 - Incorrect

$result = $query->getQuery()->getResult();
print_r($result[0]['amount']); // 553.47
?>

Sql for the above:

SELECT DISTINCT id0 FROM (SELECT q0_.id AS id0, SUM(q0_.price) AS sclr36 FROM Q q0_ WHERE q0_.id IN (19, 20, 22) GROUP BY q0_.customer_id) dctrn_result LIMIT 30 OFFSET 0
SELECT q0_.id AS id0, SUM(q0_.price) AS sclr36 FROM Q q0_ WHERE q0_.id IN (19, 20, 22) AND q0_.id IN ('19') GROUP BY q0_.customer_id
SELECT q0_.id AS id21, SUM(q0_.price) AS sclr36 FROM Q q0_ WHERE q0_.id IN (19, 20, 22) GROUP BY q0_.customer_id

Sql with fetchJoin = false (new DoctrineORMAdapter($query, false))

SELECT q0_.id AS id0, SUM(q0_.price) AS sclr36 FROM Quote q0_ WHERE q0_.id IN (19, 20, 22) GROUP BY q0_.customer_id LIMIT 30 OFFSET 0
SELECT q0_.id AS id0, SUM(q0_.price) AS sclr36 FROM Quote q0_ WHERE q0_.id IN (19, 20, 22) GROUP BY q0_.customer_id

As far as I know, docs don't explain that it is up to the developer to keep the object graph consistent instead of relying on Doctrine ORM for everything.

For example, for many to many, examples like following may be used:
https://gist.github.com/3121916

When creating a new Post and adding it to a collection in an existing Thread (i.e. loaded from the database), referencing Thread's posts collection in Post's @PostPersist callback returns the Post twice. To clarify, this only happens when Thread was previously persisted. If I'm creating a new Thread object the code works as expected. I've included some sample code to better illustrate my issue.

I don't know if this is a bug, or if I'm doing something that I shouldn't be, but I couldn't find this limitation mentioned in the documentation, and this seems to go against the expected behavior.

Here are my sample entities:

/**
 * @Entity
 * @Table(name="thread")
 */
class Thread
{
    /** 
     * @Id
     * @GeneratedValue
     * @Column(type="integer")
     */
    protected $id;

    /** 
     * @OneToMany(targetEntity="Post", mappedBy="thread", cascade={"persist", "remove"})
     */
    protected $posts;

    public function __construct() {
        $this->posts = new ArrayCollection();
    }   

    public function getPosts() {
        return $this->posts->toArray();
    }   

    public function addPost(Post $post) {
        $post->setThread($this);
        $this->posts->add($post);
    }   
}

/**
 * @Entity
 * @Table(name="post")
 * @HasLifecycleCallbacks
 */
class Post
{
    /** 
     * @Id
     * @GeneratedValue
     * @Column(type="integer")
     */
    protected $id;

    /** 
     * @ManyToOne(targetEntity="Thread", inversedBy="posts")
     * @JoinColumn(name="thread_id", referencedColumnName="id")
     */
    protected $thread;

    public function getId() {
        return $this->id;
    }   

    /** 
     * @PostPersist
     */
    public function onPostPersist() {
        $posts = $this->thread->getPosts();
        foreach ($posts as $post) {
            echo 'id: ' . $post->getId() . ' type: ' . get_class($alert) . '<br />';
        }   
    }

    public function setThread(Thread $thread) {
        $this->thread = $thread;
    }   
}

And the calling code:

// Grab an existing thread.
$thread = $em->getReference('Thread', 1); 
$thread->addPost(new Post());
$em->flush();

This outputs:

id: 1 type: Post
id: 1 type: Post

Alternatively:

// Create a new thread.
$thread = new Thread()
$thread->addPost(new Post());
$em->persist($thread);
$em->flush();

This outputs:

id: 1 type: Post

It would be nice if the Doctrine caching drivers would support the eAccelerator library.

This issue is created automatically through a Github pull request on behalf of jappie:

Url: https://github.com/doctrine/doctrine2/pull/512

Message:

This method behaves like setParameters() before version 2.3:
It will add new parameters to the collection, and override any existing positions/names.

It can take a Doctrine\Common\Collections\ArrayCollection with Doctrine\ORM\Query\Parameter objects, as well as a plain array with key/value pairs, as argument.

This will greatly ease the upgrade to Doctrine 2.3, because you only need to perform a project-wide replace of setParameters with addParameters, in stead of going into your code and determine if calls to setParameters are ok or need refactoring.

I've also added unit-tests to maintain integrity.

Today I spent some time solving a PHP 'White Screen of Death'. I traced it back to a Entity of which the proxy's __load() function was invoked because af a EXTRA_LAZY association. Due to incorrect database contents (the entry ID was changed due to an update: referential integrity broke), the __load() query resulted in no results. The EntityNotFoundException did for some reason not show up in our logs, probably because the lazy load was triggered by a magic __toString() function.

The cause is because of the way we populate or tables with domain data:

SET FOREIGN_KEY_CHECKS = 0;
#IMPORT STUFF from CSV
SET FOREIGN_KEY_CHECKS = 1;

MySQL does not trigger any errors when the foreign key checks are turned back on, leaving the table in an inconsistent state.

To prevent this, I found some information in this post: http://www.mysqlperformanceblog.com/2011/11/18/eventual-consistency-in-mysql/, which I used to come with the following queries

#Check the constraints of a specific database
SELECT *
	FROM INFORMATION_SCHEMA.KEY_COLUMN_USAGE 
		WHERE TABLE_SCHEMA = 'databaseName'
		AND REFERENCED_TABLE_NAME IS NOT NULL

#Generate table specific queries to find orphaned entries
SELECT CONCAT(
	 'SELECT ', GROUP_CONCAT(DISTINCT CONCAT(K.CONSTRAINT_NAME, '.', P.COLUMN_NAME,
	  ' AS `', P.TABLE_SCHEMA, '.', P.TABLE_NAME, '.', P.COLUMN_NAME, '`') ORDER BY P.ORDINAL_POSITION), ' ',
	 	'FROM ', K.TABLE_SCHEMA, '.', K.TABLE_NAME, ' AS ', K.CONSTRAINT_NAME, ' ',
	 		'LEFT OUTER JOIN ', K.REFERENCED_TABLE_SCHEMA, '.', K.REFERENCED_TABLE_NAME, ' AS ', K.REFERENCED_TABLE_NAME, ' ',
	 		' ON (', GROUP_CONCAT(CONCAT(K.CONSTRAINT_NAME, '.', K.COLUMN_NAME) ORDER BY K.ORDINAL_POSITION),
	 		') = (', GROUP_CONCAT(CONCAT(K.REFERENCED_TABLE_NAME, '.', K.REFERENCED_COLUMN_NAME) ORDER BY K.ORDINAL_POSITION), ') ',
	 		'WHERE ', K.REFERENCED_TABLE_NAME, '.', K.REFERENCED_COLUMN_NAME, ' IS NULL;'
	  )
    INTO OUTFILE '/tmp/verifyDatabaseTableIntegrity.sql'
    FROM INFORMATION_SCHEMA.KEY_COLUMN_USAGE K
      INNER JOIN INFORMATION_SCHEMA.KEY_COLUMN_USAGE P
        ON (K.TABLE_SCHEMA, K.TABLE_NAME) = (P.TABLE_SCHEMA, P.TABLE_NAME)
        AND P.CONSTRAINT_NAME = 'PRIMARY'
    WHERE K.TABLE_SCHEMA = 'databaseName'
      AND K.REFERENCED_TABLE_NAME IS NOT NULL
      GROUP BY K.CONSTRAINT_NAME;
	

By running the generated queries, we can now easily find the records that break referential integrity.

It might be an idea of adding this functionality to the orm:validate-schema, or a new orm:validate-database-integrity?

It would be nice if supported writing in DQL:

SELECT LOWER(a.name) AS name FROM User a WHERE name LIKE ?

The resulting sql:

SELECT LOWER(c0_.name) AS sclr0 FROM users c0_ WHERE LOWER(c0_.name) LIKE ?

Currently, using MEMBER OF for a ManyToMany collection does a join on the table of the related entity, whereas all it needs is in the join table.

Using the following DQL:

SELECT p FROM Player p
WHERE NOT :team MEMBER OF p.targetedBy

Here is the current generated SQL:

WHERE NOT EXISTS (SELECT 1 FROM player_team p1_ INNER JOIN Team t2_ ON p1_.team_id = t2_.id WHERE p1_.player_id = p0_.id AND t2_.id = ?)

whereas it could drop the join:

WHERE NOT EXISTS (SELECT 1 FROM player_team p1_ WHERE p1_.player_id = p0_.id AND p1_.team_id = ?)

This issue is created automatically through a Github pull request on behalf of Powerhamster:

Url: https://github.com/doctrine/doctrine2/pull/460

Message:

Mapping drivers now use extended classmetadata class to find constants of generator types.
Method completeIdGeneratorMapping is now protected and can be extended

XML mapping :

<id name="id" type="string" length="16"/>

Generate SQL :

id varchar(255) not null

It's not possible with XML mapping to have :

 id varchar(16) not null

Because tag "id" doesn't support "length" attribute.
Please add this attribute

Hi!
Currently i'm heavy using doctrine to generate entities starting form database schema (aprox 500 tables with thousand of relations).

I'm trying to detect some inheritance cases, but there is a problem.

Doctrine always uses PHP class inheritance to detect entity hierarchy, but generating entities starting from database, i have not yet any php class.

There is a better way to detect entities hierarchy? Without php classes...
Mapping files should be self-sufficient, even without php files.

The practical case is:
in DatabaseDriver i'm trying to call $metadata->addDiscriminatorMapClass($name, $className) method, but it raises an exception if $classNam does not exists.

Even if i manualy create XML mapping files, and then i try to generate php entityes. There is the same problem. XmlDriver tries to call setDiscriminatorMap method that raises the same exception.

Hi,

I'm working with Symfony 2.1, and I need to know if an association is nullable for a given entity (to know if a form field should be marked as 'required'). So I'd like to have a isAssociationNullable() method in the ClassMetadataInfo class, that should do the same thing that the isNullable() method does for fields.

You can see more information about the problem on the Symfony issue.

Thanks.

It would be really nice to have @Any and @ManyToAny relations/annotations implemented like on Hibernate.
http://docs.jboss.org/hibernate/orm/4.1/javadocs/org/hibernate/annotations/ManyToAny.html
Right now I've implemented these in a Symfony2 bundle (that I'd be happy to share once it's ready and a bit documented), using listeners on postLoad, preFlush and prePersist
However I think this is a very common use case that anyone will encounter at least once/twice in every middle/big-sized project, and for this reason I think this should be implemented as a core feature.

The code example here:
http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/tools.html#configuration-non-pear

In the second "use" statement it references a "EntityManagerHelper" from the "Doctrine\DBAL\Tools\Console\Helper\" package. However, it does not exist there. It does exist in the "Doctrine\ORM\Tools\Console\Helper\" package though, and replacing it seems to work.

The SQLFilter currently only accepts string parameters which would result in SQL like:

"tableAlias.column = '$filterParameter'"

To filter an Entity that has a lifecycle, this can be usefull to filter Entities that are in a specific state, for example:

"tableAlias.state = 1"

To be able to apply the filter on an Entity that can be in multiple states, it is usefull to be able to assign an array of states using setParameter:

$allowedStates = array(1,2,3,4);
$filter->setParameter('allowedStatesParam', $allowedStates);
sprintf("tableAlias.state IN (%s)", implode(',', $this->getParameter('allowedStatesParam')));

to eventually result in:

"tableAlias.state IN (1,2,3,4)"

However, this is currently not supported, it seems to go wrong on the PDO::quote() of the parameter. The SQL works ok when setting it statically in the filter, not taking the parameter into account.

It would be nice to have support for arrays on the setParameter()

When applying both @Column and @ManyToOne annotations to a field, it blows up with crazy internal errors. It would be great if this case – and similar cases – could throw a nice exception which tells the user what he did wrong.

Hello,

Reference Guide topic 4.8 Identifers/Primary Keys(http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/basic-mapping.html#identifiers-primary-keys) states that: "Every entity class needs an identifier/primary key."

However, example in topic 7.3. Class Table Inheritance (http://docs.doctrine-project.org/projects/doctrine-orm/en/2.1/reference/inheritance-mapping.html#class-table-inheritance) doesn't contain any definitions for Ids.

Consider following code:

/**
 * @Entity
 * @InheritanceType("JOINED")
 * @DiscriminatorColumn(name="discr", type="string")
 * @DiscriminatorMap({"person" = "Person", "employee" = "Employee"})
 */
class Person
{
    /** @Id @Column(type="integer") */
    private $id;

    public function getId() {
        return $this->id;
    }
}

/** @Entity */
class Employee extends Person
{
    /** @Id @Column(type="integer") */
    private $id;

    // Overrides parent to retrieve private
    public function getId() {
        return $this->id;
    }
}

// create instances and $em->persist(...)

// $person instanceof Person
$person->getId(); // Returns id.

// $employee instanceof Employee
$employee->getId(); // Returns null. Private $id in subclass isn't assigned.

Please clarify correct use of identifiers in CTI subclass entities. Should subclasses contain any definitions of identifiers?

Thank you!

Ludek

Hi all,

in a big project if you have queries spread out in different
repositories,
when you have to modify a cache lifetime, you have to search the query and
modify the code, than test it.
Is not so easy also to answer to 'how much is the cache for the query XYZ?'

the idea:
Each group of repository (bundle) should have in a single point maybe into its config file a place where you could set the lifetime of the various queries.

see the code for a better explanation
https://gist.github.com/3075742

the pro: a better handling of the cache mechanism
cons: ?

Do you think is a good approach?
Have you ever had a similar problem?

Thanks

liuggio

I'm not sure if this is really possible..

but to improve performance we should consider generate custom entity persisters.

Now entity persister are not cached,
if we generate it, we can create performance improvement in hidrators, avoiding checks and sql generation every time that an persister is called.

To improve performance we should consider generate custom hydrators per entity or per rsm.

I think that will create a huge performance improvement, avoiding checks every time that an entity/result is hydrated.

Decimal type in mysql converts as a string in php. This is great as Decimal has a much higher precision than a float or double and that precision would be lost if converted to a float in PHP. Fine! But when doing calculations (as my numbers do not require an enormous precision gmp_ functions are not necessary) php converts these strings into floats. Then, when computing the changeset, as the value is compared with === is marked as a change even though there is none ("5.00" string vs 5.00 float) and an UPDATE for that row is made. Would it be possible to check only for simple equality "==" instead of type equality "===" when dealing with scalar types?

Another example of this would be the boolean type, that it is stored as an integer 1 in mysql but converted to a boolean true in php. If during the execution of my code that boolean gets converted to an integer 1, that will trigger an UPDATE also because 1 !== true.

Should this be my responsability or doctrine should be a little more flexible regarding comparisons? Thanks!!

When a trait with included setters and getters is used and generate entities is called, doctrine add another set of getters and setters to the "main" entity where the trait is used.

Currently Doctrine\ORM\Query\ResultSetMapping can only be used to query the database for entities using the EntityManager::createNativeQuery method. It would be great if we could use this as well for objects that are not entities. That way we can create simple DTO's and map them to a query using the ResultSetMapping.

I'll open a PR If there are no objections.

When using a LIKE condition in a WHERE clause, I would like to be able to use the foreign key of a single value association. For example:

Imagine Product and Department are both entities. Department has the field Name with a unique index. Product is has a ManyToOne association called Department with the Department entity, referencing the field Name.

I would like to write:
SELECT FROM Product p WHERE p.Department LIKE '% Tools'

However, at present I need to write:
SELECT FROM Product p
JOIN p.Department d
WHERE d.name LIKE '% Tools'

The issue is one of performance. On large record sets the first query runs several magnatudes more quickly than the second, particularly when four or five joins are involved.

Looking at the DQL grammar from the online docs, the relveant lines are:

LikeExpression ::= StringExpression ["NOT"] "LIKE" string ["ESCAPE" char]
StringExpression ::= StringPrimary | "(" Subselect ")"
StringPrimary ::= StateFieldPathExpression | string | InputParameter | FunctionsReturningStrings | AggregateExpression

The problem is the StateFieldPathExpression in StringPrimary. When used with a LikeExpression a SingleValuedAssociationField should be allowed also.

It is too hard handle rows duplication now. As adviced here http://stackoverflow.com/questions/3967226/checking-for-duplicate-keys-with-doctrine-2 we should catch \PDOException with code 23000. This is bad behavior.

So I propose:
a. Add DuplicateRowException wich will throwed by flush method (maybe wrapped PDOException) with method "getDuplicatedProperties" (based on sql error description parsing + mapping)
b. Add method findOneOrCreate (hello, Propel!) or updateIfExists to ObjectInterface, EntityManager
c. Combine a, b

PDO supports grouping by a field when doing a fetchAll() by using PDO::FETCH_GROUP. This can in many places be useful. http://blog.stealth35.com/2011/08/17/pdo-fetch-group.html desribes the behavior quite well.

8. Working with Objects:
"... if the relationships from X to these other entities are mapped with cascade=PERSIST or cascade=ALL (see "Transitive Persistence")."
"... with cascade=REMOVE or cascade=ALL (see "Transitive Persistence")."
and so on.
Maybe it's reasonable to make "Transitive Persistence" as links to http://www.doctrine-project.org/docs/orm/2.1/en/reference/working-with-associations.html#transitive-persistence-cascade-operations

Hi.
It would be nice if the set-methods of the generated entity classes would return the entity instance itself, so that method chaining is possible.
Example:

$user = new User();
$user->setUsername()
     ->setFirstname()
     ->setEmail();

In PHP 5.4 we can do even more nicer:

$user = new User()
     ->setUsername()
     ->setFirstname()
     ->setEmail();

If this is not wanted by everyone, the console tool could get a new argument to define if method chaining should be used or not.

Implementation of this improvement would be very nice. Thanks.

I am refering to the examples on

http://www.doctrine-project.org/docs/orm/2.0/en/reference/batch-processing.html

Let's say you want to process a bulk of 25 objects and have a batchsize of 20.
With the code provided the last 5 would not be saved as far as I understand unless you do another flush after the for-loop.

This is probably very clear to any experienced Doctrine developer but maybe it is also confusing for beginners like me (some internet sources say, that flush is executed automatically at the end of the request, but obviously it is not). Maybe this could be mentioned somewhere?

I am running a query that JOINs three tables, with a simple WHERE:

$q = $em->createQuery("

SELECT cat, n, c
FROM Project_Model_NoticeCategory cat
JOIN cat.notices n
JOIN n.chapters c
WHERE
c.id = :chapter_id

");

When I do this:

  $q->setParameter('chapter_id', 1);
  $a = $q->getResult();

  $q->setParameter('chapter_id', 2);
  $b = $q->getResult();

$b always has the wrong results. Running the following code:

  $q->setParameter('chapter_id', 1);
  $a = $q->getResult();

  $q->setParameter('chapter_id', 2);
  $b = $q->getResult();
  $z = $q->getArrayResult();

BUG Results: $b != $z (getArrayResult IS CORRECT, it refreshes the results) Note: $a==$b (which is wrong)

Explanation:

There is a chapter table, this has a many-to-many join to notices (these are meta info
about the chapter – a little like tagging a blog post) the notices are grouped into
categories.

Data model:

/**
 * @Entity
 * @Table(name="chapter")
 */
class Project_Model_Chapter
{
    /**
     * @Id @Column(type="integer")
     * @GeneratedValue(strategy="AUTO")
     */
    private $id;
 
    /** @Column(type="string") */
    private $title;

	/**
	 * @ManyToMany(targetEntity="Project_Model_Notice", mappedBy="chapters")
	 */
	private $notices;
	
	.... /lots of code snipped/ ....
	
}


/**
 * @Entity
 * @Table(name="notice")
 */
class Project_Model_Notice
{
	/**
     * @Id @Column(type="integer")
     * @GeneratedValue(strategy="AUTO")
     */
    private $id;
 
    /** @Column(type="string") */
    private $title;
	
	/**
	 * @ManyToMany(targetEntity="Project_Model_Chapter", inversedBy="notices")
	 * @JoinTable(name="chapter_notice")
	 */
	private $chapters;
	
	/**
	 * @ManyToOne(targetEntity="Project_Model_NoticeCategory", inversedBy="notices")
	 */
	private $notice_category;
	
	.... /lots of code snipped/ ....
	
}

/**
 * @Entity
 * @Table(name="notice_category")
 */
class Project_Model_NoticeCategory
{
    /**
     * @Id @Column(type="integer")
     * @GeneratedValue(strategy="AUTO")
     */
    private $id;
	/** @Column(type="string") */
    private $title;
	
	/**
	 * Bidirectional - One-To-Many (INVERSE SIDE)
	 *
	 * @OneToMany(targetEntity="Project_Model_Notice", mappedBy="notice_category", cascade={"persist", "remove"})
	 */
	private $notices;

	.... /lots of code snipped/ ....
	
}

Data fixtures:

$tools = new \Project_Model_NoticeCategory;
$tools->setTitle('Tools');
		
$spanner = new \Project_Model_Notice;
$spanner->setTitle('spanner');
$tools->addNotice($spanner);
		
$drill = new \Project_Model_Notice;
$drill->setTitle('power drill');
$tools->addNotice($drill);
		
$this->em->persist($tools);
$this->em->flush();

$tools = new \Project_Model_NoticeCategory;
$tools->setTitle('Safety');
		
$gloves = new \Project_Model_Notice;
$gloves->setTitle('gloves');
$tools->addNotice($gloves);
		
$goggles = new \Project_Model_Notice;
$goggles->setTitle('goggles');
$tools->addNotice($goggles);
		
$this->em->persist($tools);
$this->em->flush();

$chapter1 = new \Project_Model_Chapter;
$chapter1->setTitle('Chapter 1');
$this->em->persist($chapter1);

$chapter2 = new \Project_Model_Chapter;
$chapter2->setTitle('Chapter 2');
$this->em->persist($chapter2);

$chapter1->addNotice($spanner);
$chapter1->addNotice($gloves);

$chapter2->addNotice($spanner);
$chapter2->addNotice($gloves);
$chapter2->addNotice($drill);
$chapter2->addNotice($goggles);

// now persist and flush everything

Initial investigation:

I think it has something to do with HINT_REFRESH ? Stepping through:

ObjectHydrator->_hydrateRow
ObjectHydrator->_getEntity

when it requests the Project_Model_Category from the unit of work, it
seems that the second query is simply grabbing the cached results from
the first results. This MUST be wrong as the second query uses a
different query (the ID changes) and all the results are wrong.

Instead of this piece of code:

$literal = $expr->literal($v);
$expr->andX(
  $expr->eq('at.key', $expr->literal($k)),
  $expr->orX(
    $expr->eq('a.valueInt', $literal),
    $expr->eq('a.valueText', $literal),
    $expr->eq('a.valueDate', $literal)
  )
);

I would like to simplify my query by using this syntax:

$expr->andX(
  $expr->eq('at.key', $expr->literal($k)),
  $expr->in($expr->literal($v), array('a.valueInt', 'a.valueText', 'a.valueDate'))
);

I am using a string field for data and for an optional relation too.
I am not using a ID because the second entity is from a third party application.

I used two variables in my entity mapping to the same field.

/**

• @var string $an
  *
• @ORM\Column(name="an", type="string", length=20, nullable=false)
  */
  private $an;

/**

• @ORM\OneToOne(targetEntity="DataLinked")
• @ORM\JoinColumn(name="an", referencedColumnName="part")
  */
  private $linked;

The getter is working fine.

The problem occurs when I create a new entity and would like to persist it.
As the field is used twice, the value of the second variable is erasing the first value.

At the line 525 of Doctrine\ORM\Persisters\BasicEntityPersister , I added the following test to update a null value only if there is no fieldName existing.

...
foreach ($assoc['sourceToTargetKeyColumns'] as $sourceColumn => $targetColumn) {
if ($newVal === null) {
if(!isset($this->_class->fieldNames[$sourceColumn]) || in_array($sourceColumn, $this->_class->identifier))

} else if ($targetClass->containsForeignIdentifier) {
...

(!isset($this->_class->fieldNames[$sourceColumn]) : Test if there is no existing fieldName
in_array($sourceColumn, $this->_class->identifier)) : avoid skipping identifier definition because ID is listed in fieldNames!

What do you think about that?

Thanks.

We should be able to tell Doctrine that we want the result of requests on ReadOnly marked Entities to be cached.

For instance:
$person->getMoodInformation();
$person->getCityInformation();

CityInformation are not likely to change so it would make sense to cache it and retrieve only MoodInformation (by using an annotation on the concerned Entity).

What would be even better is to tag which properies we want to hydrate from database and which properties we want to hydrate from cache.

It would be nice if the command line orm schema-tool would suggest an index on the discriminator column for single inheritance tables. Since that column would almost always be in the query, I can't think of a case when you wouldn't want it to be in an index

404 on Common download link : http://www.doctrine-project.org/projects/common/download

The Doctrine 2 reference documentation defines the comparison operators syntactically, as:

ComparisonOperator  ::= "=" | "<" | "<=" | "<>" | ">" | ">=" | "!="

Now, without experimentation, it is pure guesswork to tell whether e. g. null = null, or e. g. null <> null, or null = 0, or null <> 0 are considered true statements in DQL.

In SQL semantics, all four statements would be false (or more precisely, null). In PHP semantics, both null == null and null == 0 would be true, while null != null and null != 0 would be false.

It would be helpful to have the semantics of the comparison operators defined. While comparisons with non-null values behave in a common-sense way, it is hard to guess how queries involving comparison operators on fields allowing null values or null query arguments will filter the results, without knowing the exact semantics of the comparison operators with regard to null values.

It would be great if this could be clarified in the docs.

Thanks.

Hi there,
AbsractFileDriver is using the filename to know the managed class.

It's a cool feature because it's allow loading on-demand.
The problem is, that the filename must be the name of the Class.

It should be great to be able to manually map XML/YAML File description to a Class, like :
$drivers->addMappingFile ( array ( "filename" => "class", "filename2" => "class2") );

This feature is simple to implement, just add a new array inside AbsractFileDriver to know the mapping.
When using the current method with addPaths, parse the folder to get traditional XML/YAML file where filename corresponding to classname and add it to the mapping array.

AbsractFileDriver->getAllClassNames () just return value of mapping array.
The mapping array is store inside cache.

With this new feature, it allow developers to create a pretty folder that contains entities mapping.

Armetiz.

Currently we have a set of Lifecycle events, but they seem to be misleading both in actual implementation and documentation.

One good example is prePersist and postPersist, which is only fired when you're creating new entities. It should be renamed to preInsert and postInsert.
As of preUpdate and postUpdate, they seem quite valid.

But if we rename prePersist and postPersist to (pre|post)Insert, we may have a situation where you wanna cover both insert and update.
For this, (pre|post)Persist should be reinstated, but acting differently from what it does currently.

My tweet:

> @beberlei Have you heard of overriding the ProxyFactory to allow caching of lazy-loaded entities? Trying to do that now

The majority of our data is quite stagnant and so I was shoehorning the capability of the generated proxies to use a custom class.

My new proxy, in short, will lazy-load the data as normal the first time around, but also stores it in Memcache using an injected adapter. Upon subsequent lazy-loading, memcache is used rather than a call to the DB.

I can't decide if this is better suited for the EntityPersister (which has already been discussed at length), but it seems to fits nicely with a custom proxy.

The Documentation for @JoinColumn annotation states:
"This annotation is not required. If its not specified the attributes name and referencedColumnName are inferred from the table and primary key names."

However, this seems not to be correct. If you have non-standard name for the @Id columns for a @OneToMany/@ManyToMany the name and referencedColumnName are not correctly inferred.

https://gist.github.com/e61bf8f4462870ffd4f3

Using the following code, when you go from (1, 2) to (1), (2) is deleted as expected. However, if you go from (1, 2) to (2), (1) and (2) are deleted and (2) is then inserted. Is this the desired behaviour? (i.e. 2 extra queries)

$bracket->getTournamentLocations()->takeSnapshot();

$col = $bracket->getTournamentLocations()->unwrap();

$col->clear();

foreach ($form->getValue('tournamentLocations') as $id) {
    $col->add($em->getReference('Tournaments_Model_TournamentLocation', $id));
}

$bracket->getTournamentLocations()->setDirty(true);

Please make it more clear that prePersist and postPersist events are called only when creating new entity (that is, prior and after a database insert).

IRC log:
helmer
Hi. I have a question regarding (pre|post)persist events. Why are the events ignored for second persist (should fire prePersist) + flush (should fire postPersist) in the following pastie: http://pastebin.com/V8CrPWkM Is it a bug or am I missing sth?
Stof
helmer: there is no second persist. persist() means saying Doctrine to manage the entity. Once it is managed, you are not persisting it anymore but updating it
helmer
Stof: so basically one could define these two events for themselves as (pre|post)Insert?
beberlei
yes
helmer
thanks beberlei&stof! though a suggestion to docteam .. perhaps make it more clear to people like me, current doc can be kind of misleading ie: "There are two ways for the prePersist event to be triggered. One is obviously when you call EntityManager#persist()" http://www.doctrine-project.org/docs/orm/2.0/en/reference/events.html#prepersist

Pastie code (probably expired by now):
$entity = new Entity();

$entity->setSomething('xxx');
$this->em->persist($entity);
$this->em->flush();

$entity->setSomething('yyy');
$this->em->persist($entity);
$this->em->flush();

Hi,
I've been having issues with AnnotationDriver crawling in my directories and loading files.
I have a few classes that require specific libraries loaded, and I don't want the AnnotationDriver to load them.

For example, I have my descendant of PHPUnit_Framework_TestCase in libs and the driver just dies, because PHPUnit is not loaded, and I don't want to load it, to be able to finish the process.

Solution would be add method AnnotationDriver::addExcludePath, whose name speaks for itself

Temporarily, I had to extend the AnnotationDriver and overload the crawling process, which is realy annoing, because I had to copy the whole method with all its exceptions and I would have to maintain it, till this will be in Doctrine. Can be viewed here https://github.com/Kdyby/Framework/blob/master/libs/Kdyby/Doctrine/Mapping/Driver/AnnotationDriver.php

Thanks
Filip

I suggest adding a new feature to SchemaTool, which allows you to ignore tables, which contain unsupported column types.

Use case:

• You have a legacy database, or a database that is also shared with another application
• You want to use SchemaTool to speed up development
• The database contains tables which are not used in the Doctrine 2 application, and contain unsupported column types

I've encountered this already a few times myself - Basically if you try to use orm:schema-tool:update with a database that contains tables with unsupported column types, it'll throw an error and you won't be able to use it at all. Because schema tool is extermely convenient when developing, I think it would be very useful to have support for this feature.

Implementation:

I think this should be doable by just changing SchemaTool/SchemaManager so, that SchemaManager would contain an additional method (or flag) which works like createSchema, but ignores tables that cause an exception, and SchemaTool would include a flag for using this instead of the standard approach.

I'm looking into implementing this myself, and will submit a patch if this seems like a reasonable approach.

If an IDE would like to support annotations, it's currently only possible to display all resolvable classes in a code-hint menu when autocompleting annotations, as
basically any class can be used for annotations.

To make it possible for IDEs to detect classes which are explicitly meant to be used as annotations, it would be nice to agree on some common
way of documenting annotations in PHPDocBlocks.

Here's an example of what this could look like:

https://github.com/pulse00/doctrine2/commit/25a14e9edc406edfd33e54fc38922a191e9cbe83

This way IDEs can prioritize annotated classes in code-hints and add additional information to the user.

The error message you get when an entity is missing an ID when attempting to persist it is rather confusing.

"Entity of type Some\Entity\Name is missing an assigned ID."

This does not tell me anything at all. I had absolutely no idea what an assigned ID was. I totally randomly noticed that I had mistyped @GeneratedValue as @GenratedValue, and fixing it fixed the issue.

Perhaps the message makes sense if you're familiar with Doctrine 2 internals, but I think it should be changed to something more helpful, such as "Entity of type X is missing primary key".

Alternatively it could keep the same message, but it could suggest a possible error ("Does the entity have a primary key set?") or perhaps the documentation could include a reference to it to help debug.

An event triggered when loading collections would be useful for performances. The use case would be batch querying some stuff instead of doing a query per object of the collection in a postLoad event.
For instance, the Translatable extension from https://github.com/l3pp4rd/DoctrineExtensions loads the translations on postLoad which result in many queries. Being able to load them all in a single query would be useful.

With the schema:

 
Image
    @Id
    $id

Tag
    @Id
    $Id

Tag_Image
    @Id
    @OneToOne(targetEntity="Tag")
    @JoinColumn(name="tag")
    $tag

    @Id
    @OneToOne(targetEntity="Image")
    @JoinColumn(name="image")
    $image

Given the following DQL,

    SELECT img
    FROM Image 
    LEFT JOIN img.tags tag
    WHERE tag.id=:tag

Doctrine Generates this SQL

 
    SELECT i0_.id AS id1 
    FROM Image i0_ 
    LEFT JOIN Tag_Image t2_ 
        ON i0_.id = t2_.image 
    LEFT JOIN Tag t1_ 
        ON t1_.id = t2_.tag 
    WHERE t1_.id = 37

Which unncessarily joins against Tag, given that the foreign key Tag.id is also found in Tag_Image.tag.

The XML Mapping documentation indicates that "orphan-removal" is an attribute on One-to-One or Many-to-One elements, but in DoctrineORM version 2.0 that does not work. It seems I have to make "orphan-removal" a child attribute of those elements to turn it on.

Just to be clear:

Documentation says to do this, but it does not work:
<one-to-many field="..." orphan-removal="true"> ... </one-to-many>

This does work:
<one-to-many field="..."> <orphan-removal>true</orphan-removal> </one-to-many>

Since DDC-514 there is a default in place for this, thus the description is wrong in stating this is required.

I registered an event listener to my entity manager and on a postLoad event, I want to prepare some data in a nice way (fetch translations for my library + store into associative array into entity). Here's my snippet:

class TranslationListener implements EventSubscriber
{
    public function getSubscribedEvents()
    {
        return array(Events::postLoad);
    }

    public function postLoad(LifecycleEventArgs $args)
    {
        $em = $args->getEntityManager();
        $entity = $args->getEntity();

        if ($entity instanceof Lib) {
            $qb = $em->createQueryBuilder();
            $qb = $qb->select('T')->from('Translate', 'T')->join('T.locale', 'TT')->where('T.lib = ?1')->setParameter(1, $entity->idLib);
            $res = $qb->getQuery()->getResult();
            foreach ($res as $tr) {
                $entity->tr[$tr->locale->idLocale] = $tr;
            }
        }
    }
}

When this code is run (eg. getting the Library objects), I got a crash where getResult() is called:

Fatal error: Call to a member function fetch() on a non-object in /home/thepianoguy/testproject/trunk/lib/Doctrine/ORM/Internal/Hydration/ObjectHydrator.php on line 126
Call Stack
#	Time	Memory	Function	Location
1	0.0004	656080	{main}( )	../index.php:0
2	0.1081	18760176	TApplication->run( )	../index.php:48
3	0.2637	33817288	TApplication->runService( )	../TApplication.php:382
4	0.2637	33817288	TPageService->run( )	../TApplication.php:1095
5	0.2698	34788448	TPageService->runPage( )	../TPageService.php:444
6	0.2715	34986768	TPage->run( )	../TPageService.php:498
7	0.2716	34989128	TPage->processNormalRequest( )	../TPage.php:198
8	0.3383	42770128	TControl->loadRecursive( )	../TPage.php:215
9	0.3383	42770208	ContactUserAddEdit->onLoad( )	../TControl.php:1286
10	0.3383	42771912	ContactUserAddEdit->loadData( )	../ContactUserAddEdit.php:56
11	0.3452	43436904	Doctrine\ORM\AbstractQuery->getResult( )	../ContactUserAddEdit.php:124
12	0.3452	43437296	Doctrine\ORM\AbstractQuery->execute( )	../AbstractQuery.php:366
13	0.4160	47009328	Doctrine\ORM\Internal\Hydration\AbstractHydrator->hydrateAll( )	../AbstractQuery.php:537
14	0.4160	47011504	Doctrine\ORM\Internal\Hydration\ObjectHydrator->_hydrateAll( )	../AbstractHydrator.php:99

If I comment the line 137 in Doctrine/ORM/Internal/Hydration/AbstractHydrator in _cleanup(), my code works fine:
// $this->_stmt = null;

I think there is a problem when using alredy used entity manager and query builder inside the postLoad event.

http://www.doctrine-project.org/docs/orm/2.0/en/reference/events.html

The above URL has an example of how to register lifecycle callbacks from YAML, but actually it does not work. The correct way of doing it is mentioned on the page: http://www.doctrine-project.org/docs/orm/2.0/en/reference/yaml-mapping.html

I read mini tutorial http://www.doctrine-project.org/docs/orm/2.0/en/reference/introduction.html#mini-tutorial and try run command "doctrine orm:schema-tool --drop"