Ok, so the object relations go like this:

• Comparison
  • [1:N] Product (FK:comparison_id)
    • [1:N] Rules (FK:product_id, FK:option_id)
    • [1:N] Parameters (FK:product_id)
      • [1:N] Options (FK:parameter_id)

The testing code looks like following:

== CODE START ==

$comp = new Application_Model_Comparison();

/* Filling $comp with data */

for ($i = 0; $i < 10; $i++) {

    $product = new Application_Model_Product();

    // Options referenced in Rules
    $options = array();

    for ($j = 0; $j < 10; $j++) {

        $param = new Application_Model_Parameter();

        for ($k = 0; $k < 10; $k++) {

            $option = new Application_Model_Option();

            $param->Options->add($option);

            // Register a single option for this parameter
            if (!isset($options[$j])) {
                $options[$j] = $option;
            }
        }

        $product->Parameters->add($param);
    }

    for ($j = 0; $j < 10; $j++) {
        $rule = new Application_Model_Rule();

        $rule->Option = $options[$j];
        $product->Rules->add($rule);
    }

    $comp->Products->add($product);
}

/**
 * The first save() goes nicely, all objects
 * are created (INSERTed)
 */ 

$comp->save();

// Remove every second product
$pCount = $comp->Products->count();

for ($i = 0; $i < $pCount; $i += 2) {
    $comp->Products->remove($i);
}

/**
 * Fails while trying to save
 *
 * Comparison->Product->Parameter->Option
 * INSERT ... `parameter_id` cannot be NULL
 *
 * Comparison->Product->Rule
 * INSERT ... `product_id` cannot be NULL
 */

$comp->save();

== CODE END ==

The first save() cleans up the relation information in the graph. All child objects are INSERTED instead of UPDATE.

In PHP 5.4 there is a change in the way the object references are serialized:

Quote:
"Support for object references in recursive serialize() calls
Prior to PHP 5.4, object references where not saved in recursive serialize calls."

This minor change, breaks down serialization of collections when column of type "array" is present - double serialization occurs.
I'm attaching a patch fixing the issue.

In the attached Squema run this query against postgreSQL.
(it runs ok In mysql)

$lang = 'en';
$session = 1;
$q = Doctrine_Query::create()
->from('Sys_Trace t')
->leftJoin('t.Sys_Session s')
->leftJoin('t.Translation tr WITH tr.lang = ?', array($lang))
->leftJoin('t.Sys_Oper so')
->leftJoin('so.Translation tr2 WITH tr2.lang = ?', array($lang))
>where('t.session_id = ?', array($session));

SQLSTATE[22P02]: Invalid text representation: 7 ERROR: invalid input syntax for integer: "en"

this is the my table creation .

CREATE TABLE billboard(id BIGINT AUTO_INCREMENT,title VARCHAR (255),country_id BIGINT,zone_id BIGINT,place VARCHAR(255),occassion VARCHAR(255),itinerary VARCHAR(255),created_at DATETIME NOT NULL,description TEXT NOT NULL,PRIMARY KEY(id)) ENGINE=INNODB;

public static function getInstance()

this is the my Doctrine Table

// public function getBillboardsForAUser($userId, $limit, $offset=0)
public function getBillboardsForAUser($userId,$limit,$offset=0)
{
$query = $this->createQuery('b')
->where('b.title=?',$title);
// ->where('b.owner_id = ?', $userId);
$followings = Doctrine::getTable('Follow')->getAllFollowing($userId);
foreach($followings as $following)

$query->orderBy('b.created_at DESC')
->limit($limit)
->offset($offset);
return $query->execute();

}

plz help me what is the problem.

The Timestampable Listener only sets the timestamp if the timestamp field has not been modified:

if ( ! isset($modified[$createdName])) {
  $event->getInvoker()->$createdName = $this->getTimestamp('created', $event->getInvoker()->getTable()->getConnection());
}

When saving a copy of a Doctrine_Record that doesn't already have the timestamp fields set fails to be updated, leading to integrity constraint violation ("created_at cannot be NULL"). The reason is that all unset fields in the copy are set to an instance of Doctrine_Null, which is considered to be modifed according to the condition tested for above. To fix the issue, I modified the code above to read:

if ( ! isset($modified[$createdName]) || $modified[$createdName] instanceof Doctrine_Null) {
  $event->getInvoker()->$createdName = $this->getTimestamp('created', $event->getInvoker()->getTable()->getConnection());
}

The default value for timestamp is "0000-00-00 00:00:00", so
$e = explode('T', trim($value))
should be changed to
$e = explode(' ', trim($value))

public function validate($value)
{
if (is_null($value))

$e = explode('T', trim($value));
$date = isset($e[0]) ? $e[0]:null;
$time = isset($e[1]) ? $e[1]:null;

$dateValidator = Doctrine_Validator::getValidator('date');
$timeValidator = Doctrine_Validator::getValidator('time');

if ( ! $dateValidator->validate($date))

return true;
}

Function spliti(); is deprecated.
It need to be change to (in my own opinion):

$field_array = str_ireplace(' as ', ' as ', $field_array);
$aux2 = explode(' as ', $field_array);

thnx.

Hello,

We found a bug in Doctrine1 MSSQL Connection.
When you would like to use the following functionality: find(One)By(p1,p2)
if you use the old functionality (Symfony 1.4 support it) like this: findBy("idAnddata", array("id" => ..., "date" => ..)), you got an MSSQL error, because the values wasn't changed.

Please find the patch for it, I hope it helps to you as well.

Kind regards
Peter

Adding an invalid filename to the data-load task results in purging of all the data in the database.

I am attaching a patch that checks the loaded data if there were any values actually loaded from the fixtures.

Was able to resolve this - see comment below - but still think it counts as a bug since the source of the error is so unclear

Hello,
I'm familiarizing myself with symfony at this point, but doctrine seems like a very accessible ORM tool overall. This install will also use the apostrophe plugin though that is more a client request and it is seeming to complicate a lot of issues from what I can see. Right now, I am just trying to create some db tables in schema.yml and build them with doctrine. When running $ php symfony doctrine:build-schema I get the following error: When using the attribute ATTR_AUTO_ACCESSOR_OVERRIDE you cannot use the field name "data" because it is reserved by Doctrine. You must choose another field name.

Which would be clear enough except I'm NOT using "data" as a field name in my schema file: here's what I'm using:

sfTravelLodgingLocationsType:
columns:
lodging_name:

lodging_code:

sfTravelLodgingLocations:
columns:
lodging_type_code:

name:

address:

city:

distance:

phone:

known_2b_sold_out:

relations:
Travel_Lodging_LocationsType:

I'm assuming this is a misnamed error call ... I have found a few references to that same error in other threads but none that resolve it

Using ->limit() in conjunction with many-to-many relationships with mysql leads to strange SQL generated. The condition id IS NULL is added in such case which is not correct at all.

Here's example schema

User:
  columns:
    username: { type: string(255) }
  relations:
    Operators:
      foreignAlias: Users
      class:        Operator
      refClass:     OperatorUser

Operator:
  columns:
    username: { type: string(255) }
    type:     { type: integer }


OperatorUser:
  columns:
    user_id:      { type: integer }
    operator_id:  { type: integer }
  relations:
    Operator:
      foreignAlias: OperatorUser
    User:
      foreignAlias: OperatorUser

And here's query which generates wrong SQL

Doctrine_Core::getTable('User')
  ->createQuery('User')
  ->leftJoin('User.Operators Operator')
  ->addWhere('Operator.type = ?', 1)
  ->limit(10)
  ->offset(0)
  ->execute()
;

Expected SQL generated:

SELECT u.id AS u__id, u.username AS u__username, o.id AS o__id, o.username AS o__username, o.type AS o__type
FROM user u
LEFT JOIN operator_user o2  ON (u.id = o2.user_id)
LEFT JOIN operator o        ON o.id = o2.operator_id
WHERE (o.type = '1')
LIMIT 10

Actual SQL generated:

SELECT u.id AS u__id, u.username AS u__username, o.id AS o__id, o.username AS o__username, o.type AS o__type
FROM user u
LEFT JOIN operator_user o2  ON (u.id = o2.user_id)
LEFT JOIN operator o        ON o.id = o2.operator_id
WHERE
  u.id IS NULL # is not expected
  AND (o.type = '1')
# there's no LIMIT clause

Seems like here's code which causes the bug https://github.com/doctrine/doctrine1/blob/master/lib/Doctrine/Query.php#L1307

Hi,

I have the problem, that a foreign key constraint is not created in the SQL schema. This occurs, when the primary key is not the column 'id'.

Here is an example:

User:
columns:
username:
type: string(30)
notnull: false
email:
type: string(80)
notnull: true
gender:
type: enum
values: [0,m,f]
notblank: true
notnull: true
birthday:
type: date

Address:
columns:
user_id:
type: integer(4)
unsigned: 1
notnull: true
primary: true
some_data:
type: string(100)
relations:
User:
local: user_id
foreign: id
foreignType: one

The foreign key from contacts to users is not created in der SQL schema. But if I delete the attribute 'primary' at the column 'user_id' (and a primary key 'id' will generated), everything is okay.

Can you help me please?

With kind regards
Stephan

I happen to be using Symfony 1.4.14-DEV (r33007) and am trying to setup migrations with Oracle, which is using Doctrine_Core::ATTR_QUOTE_IDENTIFIER. This issue is in core Doctrine.

To reproduce:

1. Make a change to your schema.yml
2. Create the migrations diff by executing php symfony doctrine:generate-migrations-diff
3. New file have been created at ./lib/migration/doctrine/*_version1.php
4. Try to migrate using command php symfony doctrine:migrate
5. Results in error:

ORA-00942: table or view does not exist : SELECT version FROM migration_version. Failing Query: "SELECT version FROM migration_version"

Cause: The current connection is using quoted identifiers, so the query used to create the migration_version table when migrations are first instantiated was properly created as "migration_version" (notice quotes). But the raw SQL query used in Doctrine_Migration::getCurrentVersion() is not quoting the table or column identifiers, so Oracle can't find the table.

There are several places in Doctrine_Migration where the table and column identifiers are not quoted, thus breaking migrations when using a database with strict rules on the case of identifiers.

Even though I'm developing against Oracle, this also likely affects other drivers.

This bug was introduced by the person reporting the bug #DC-592.

When trying to generate an ALTER TABLE statement with Doctrine_Core::ATTR_QUOTE_IDENTIFIER enabled, the column identifier is not quoted, and a blank identifier is instead quoted, generating the following SQL:

ALTER TABLE "mytable" MODIFY (username "" VARCHAR2(200))

The proper SQL to be generated should be:

ALTER TABLE "mytable" MODIFY ("username" VARCHAR2(200))

While trying to develop a new Symfony frontend to an existing database - whcih unfortunately contains non-ascii character names - I ran into a lot of problems where non-ascii characters had been mangled.
After installing XDebug and digging into the issue I found that the use of strtolower on the column names was the issue, since it's not safe to use on UTF-8 strings.
I replaced all calls to strtolower with mb_strtolower and UTF-8 encoding which solved my issue. I don't know if that is the correct way of doing it or if there is a better way.
I saw one other use of mb_strtolower in doctrine and it was guarded with an if function exists... Also it might be an issue in other files as well...
I provide my patch file incase it is of any use.

Hi,

Adding/dropping UNIQUE CONSTRAINT doesn't work on PostgreSQL.

I'm attaching patch for this problem.

Best regards,
Michal

Just like in Doctrine 2 (http://www.doctrine-project.org/jira/browse/DBAL-50):

The PostgreSQL database does not create indexes for foreign key columns, the user has to create them by hand. I think that indexes for foreign keys should be created automatically

$query = new Doctrine_Query();
$query->select('e.entity_name,e.entity_id,s.id,s.parent_id,e.ffc_entity_id,c.country_id,c.country_name')
//$query->select('e.entity_name,e.entity_id,s.id,s.parent_id,e.ffc_entity_id,ea.Country')
->from('Entities e')
->leftJoin('e.EntityAddresses ea ON ea.entity_id = e.entity_id AND ea.address_type ="M"')
->leftJoin('ea.Country c ON ea.country = c.country_id')
->leftJoin('e.ActiveFactories s')
->where('e.status=1');
if(!empty($alpha))

$query->andWhere("s.company_id=".$parentId)
->andWhere("e.entity_type=2")
->andWhere('s.status=1')
->groupBy('e.entity_id');

$query = new Doctrine_Query();
$query->select('e.entity_name,e.entity_id,s.id,s.parent_id,e.ffc_entity_id,c.country_id,c.country_name')
//$query->select('e.entity_name,e.entity_id,s.id,s.parent_id,e.ffc_entity_id,ea.Country')
->from('Entities e')
->leftJoin('e.EntityAddresses ea ON ea.entity_id = e.entity_id AND ea.address_type ="M"')
->leftJoin('ea.Country c ON ea.country = c.country_id')
->leftJoin('e.ActiveFactories s')
->where('e.status=1');
if(!empty($alpha))
{
$query->andWhere("e.entity_name like '".$alpha."%'");
}
$query->andWhere("s.company_id=".$parentId)
->andWhere("e.entity_type=2")
->andWhere('s.status=1')
->groupBy('e.entity_id');

With autocommit set to off in mysqld, Doctrine_Migration::setCurrentVersion() does not have any effect. This is because the method uses raw PDO calls, which are discarded without either autocommit or an explicit COMMIT;.

We patched Doctrine as in the attachment. It works for us, but may not be the best general solution.

The patch only fixes this one issue. There are likely many areas in Doctrine that rely upon autocommit behavior in MySQL. We will continue to look for them, and supply patches as we find them. However, as we are only concerned about MySQL, our solutions will probably not apply to other PDO drivers.

I noticed this issue after setting up timestampable behavior on an aliased column in a legacy table:

 
<?php

abstract class Content_Article extends Doctrine_Record
{

  public function setTableDefinition()
  {
    $this->setTableName('t_content');
    
    $this->hasColumn('id', 'integer', 11, array('primary' => true, 'autoincrement' => true));
    // ...
    $this->hasColumn('datePost as posted_at', 'timestamp');
    $this->hasColumn('dateEdit as updated_at', 'timestamp');
    // ...
    
  }
  
  public function setUp()
  { 
    // ..
    $this->actAs('Timestampable', array('created' => array( 'name' => 'datePost',
                                                            'alias' => 'posted_at'),
                                        'updated' => array( 'name' => 'dateEdit',
                                                            'alias' => 'updated_at')));
  }

}

Before I added timestampable to this model, I was setting the timestamp fields manually, which worked fine.

I had to look at the source to find the alias option in the timestampable behavior, since it does not appear to be in the 1.2 documentation. (If this issue is invalid because it's not an officially supported option, I apologize).

After I added timestampable to the model, Doctrine began throwing an exception when I tried to save a new record:

Error: Doctrine_Record_UnknownPropertyException [ 0 ]: Unknown record property / related component "datePost" on "Content_Article" ~ [...]/Doctrine-1.2.4/Doctrine/Record/Filter/Standard.php

It appears that the alias option is used when setting the table definition in the behavior template, but not used by the template's listener when creating, updating, etc.

I'm attaching a zip with a copy of the changes I made to fix this in 1.2.4 and a git patch.

If I define a base model:

Entity:
connection: other
columns:
name: {}
size: {}

and inherited model:

Box:
inheritance:
extends: Entity
type: column_aggregation

Then file with base definition of Box does not contain bindComponent sentence to bind Box model with connection specified for Entity model.

Added "Geographical" behaviour to an entity:

 
ZipCode:
  actAs: [Geographical]
  columns:
    code: string
    [...]

Using the standard Symfony-generated admin manager, when trying to insert a new record without specifying both latitude and longitude, Doctrine generates a wrong INSERT statement for PostgreSQL, including latitude and longitude as '' (empty string)

 
Doctrine_Connection->exec('INSERT INTO zip_code (hidden, code,  country_id, latitude, longitude) VALUES (?, ?, ?, ?, ?)', array('false', '20133', '1', '', ''))

This cause this error:
SQLSTATE[22P02]: Invalid text representation: 7 ERROR: invalid input syntax for type double precision: ""

Workaround

explicitly specifying float type for latitude and longitude...

 
  actAs: 
    Geographical:
      latitude: {type: float}
      longitude: {type: float}

...still generate the column correctly as double precision in the database, but will not cause any error on inserting records.

This issue is exactly the same as DC-252, but refers to PostgreSQL

if i create class called Attribute_set that extends Doctrine_Record and setTableName in setUpTableDefinintion like in documentation setting up models section, the table name appears not to be set
getTableName returns "doctrine_record_abstract"

any joins or relations fail.

if i set it in the setUp function the setTableName appear working and returns "attribute_sets" table name. This only happens with this particular class, and relations have no effect.

I attached class example i'm expiriencing problems with, that should help reproducte this issue

I have fixed this in my own version of doctrine but unfortunately I am to far diverged from the trunk to offer a patch.

here is a test case:

public function testSubqueryInWhereWithJoinAndLimit()
    {
        $q = new Doctrine_Query();
        $q->select('u.id');
        $q->from('User u');
        $q->where('u.id NOT IN (SELECT a.id FROM User u2 LEFT JOIN u2.Album a LIMIT 1)');
        $this->assertEqual($q->getSqlQuery(), 'SELECT e.id AS e__id FROM entity e WHERE (e.id NOT IN (SELECT a.id AS a__id FROM entity e2 LEFT JOIN album a ON e2.id = a.user_id WHERE (e2.type = 0) LIMIT 1) AND (e.type = 0))');
    }

To fix the issue I changed this line in Doctrine_Query as follows:

if ( ( ! empty($this->_sqlParts['limit']) || ! empty($this->_sqlParts['offset'])) && $needsSubQuery && $limitSubquery) {

=

if ( ( ! empty($this->_sqlParts['limit']) || ! empty($this->_sqlParts['offset'])) && $needsSubQuery && $limitSubquery && !$this->isSubquery()) {

Hope that helps.

Sincerely

Will Ferrer

in bindParam method there is:
switch ($type) {
case Doctrine_Core::PARAM_STR:
$oci_type = SQLT_CHR;
break;
}
I think there should be other oci_types too. I had to add:
case Doctrine_Core::PARAM_INT:
$oci_type = SQLT_INT;
because I got ORA-06502: PL/SQL: numeric or value error: character string buffer too small. while executing
$stmt->bindParam(":result", $result, Doctrine_Core::PARAM_INT);

After adding SQLT_INT everything is ok

if you set prefix to "xyz_%s" and have the model "BackgroundColor" it will become the table => "xyz_background_color"
if you have the table "xyz_background_color" with unknown model and create the the model from the table, you will get => "XyzBackgroundColor".

The fix (diff):

368a369,370
> $tablePrefix = $manager->getAttribute(Doctrine_Core::ATTR_TBLNAME_FORMAT);
>
381d382
<
385c386
< $definition['className'] = Doctrine_Inflector::classify(Doctrine_Inflector::tableize($table));
—
> $definition['className'] = Doctrine_Inflector::classify(Doctrine_Inflector::tableize(preg_replace(sprintf('/\A%s\z/', str_replace('%s', '(.*?)', $tablePrefix)), '$1', $table)));
396c397
< $class = Doctrine_Inflector::classify(Doctrine_Inflector::tableize($table));
—
> $class = Doctrine_Inflector::classify(Doctrine_Inflector::tableize(preg_replace(sprintf('/\A%s\z/', str_replace('%s', '(.*?)', $tablePrefix)), '$1', $table)));

Two typos in Doctrine files prevents usage of symfony core_compile.yml system, or any similar compiler system :

• According to its class name, "Doctrine_Validator_HtmlColor", the file "Doctrine/Validator/Htmlcolor.php", should be named "HtmlColor.php" (note the uppercase "C" of "Color")
• The php tag of the file "Doctrine/Locking/Exception.php" is uppercased. It should be "<?php" instead of "<?PHP"

With Doctrine::ATTR_QUOTE_IDENTIFIER enabled, when you launch a query with a complex having clause, Doctrine_Query_Having class does not handle it correctly.

By example, when you track the having clause interpretation:

$query->addHaving( 'SUM( IF( s.id = ? , 1 , 0 ) ) = 0' , 7 );

At some point, Doctrine_Query_Having at line 70 return something like "`s10`.`id = ?`" instead of "`s10`.`id` = ?".

I just fix it using:

return $this->query->parseClause($func);

instead of:

return $this->_parseAliases($func);

Now, the parseAliases function is not used anymore...

See patch attached...

Loops

1. We have two versions of the application on the same server.
2. Second application has an updated database. New field is added to one of the models.
3. When the second app is hit first, query is stored in APC.
4. First app finds cached query and tries to call it. Exception is thrown as it doesn't know anything about the new field yet.

Situation often happens on shared development machine when one developer adds a field but others don't have in their models yet. It also happens on staging server if it's shared with production.

I suspect it only affects queries without explicitly listed fields.

To quickly fix the issue in my symfony project I extended Doctrine_Cache_Apc to implement namespaces (https://gist.github.com/944524). More appropriate place to fix it would be Doctrine_Query_Abstract::calculateQueryCacheHash().

Prior to 5.3.6, the MySQL PDO driver ignored the character set parameter to options. Due to MySQL's C api (and MySQLND), this is required for the proper function of mysql_real_escape_string() (the C API call). Since PDO uses the mres() C call for PDO::quote(), this means that the quoted string does not take into account the connection character set.

Starting with 5.3.6, that was fixed. So now if you pass the proper character set to PDO via driver options, sql injection is impossible while using the PDO::quote() api call.

$dsn = 'mysql:dbname=INFORMATION_SCHEMA;host=127.0.0.1;charset=GBK;';
$pdo = new PDO($dsn, $user, $pass);
$pdo->exec('SET NAMES GBK');
$string = chr(0xbf) . chr(0x27) . ' OR 1 = 1; /*';
$sql = "SELECT TABLE_NAME
            FROM INFORMATION_SCHEMA.TABLES
            WHERE TABLE_NAME LIKE ".$pdo->quote($string)." LIMIT 1;";
$stmt = $pdo->query($sql);
var_dump($stmt->rowCount());

Expected Result: `int(0)`.
Actual Result: `int(1)`.

There are 2 issues to fix. First, the documentation does not indicate that you can pass the `charset` option to the MySQL Driver. This should be fixed so that users are given the proper option to set character sets.

Secondly, `Connection::setCharset()` should be modified for MySQL to throw an exception, since the character set is only safely setable using the DSN with PDO. This is a limitation of the driver and could be asked as a feature request for the PHP core. Either that, or a big warning should be put on the documentation of the API to indicate the unsafe character set change

Note that this is the same issue reported for Doctrine2 with link: http://www.doctrine-project.org/jira/browse/DBAL-111

In brief:
Doing $result1 = Doctrine_Query::create()>... followed by $result2 = Doctrine_Query::create()>... can lead to a situation where the content of $result1 has become the value in $result2.

In detail:
The attached models.yml defines two simple tables with a One-to-Many relationship; we have people and names and each person can have multiple names. The DB can be propagated along the lines of:

INSERT INTO `tblname` VALUES (1,1,'alpha'),(2,2,'beta'),(3,3,'gamma'),(4,4,'delta'),(5,5,'epsilon'),(6,1,'aleph');
INSERT INTO `tblperson` VALUES (1),(2),(3),(4),(5);

Applying the query:

$results1 = Doctrine_Query::create()
->from('Person ppa')
->innerJoin('ppa.Name n')
->where('ppa.id = ?', 1)
->andWhere('n.text = ?', 'alpha')
->execute()
->getFirst()
->Name;

and then producing output though

print 'Results (1): '.count($results1)."\n";
foreach ($results1 as $result) print $result['text'] . "\n";
print "\n\n";

produces the expected:

Results (1): 1
alpha

Doing a similarly query to a new variable:

$results2 = Doctrine_Query::create()
->from('Person ppa')
->innerJoin('ppa.Name n')
->where('ppa.id = ?', 1)
->andWhere('n.text = ?', 'aleph')
->execute()
->getFirst()
->Name;

and printing with

print 'Results (2): '.count($results2)."\n";
foreach ($results2 as $result) print $result['text'] . "\n";
print "\n\n";

produces the expected:

Results (2): 1
aleph

but printing out the first result object again at this point gives:

Results (1): 1
aleph

which is unexpected - "aleph" rather than "alpha".

If, the second query was altered to

->where('ppa.id = ?', 2)
->andWhere('n.text = ?', 'beta')

then all three output results are as expected.

test.zip contains corresponding test files.

When creating an UPDATE query, the table names are not aliased like in a SELECT statement. This causes ambiguous column names when JOINING in an UPDATE.

E.g.
$q = $this->createQuery('st')
->update('SomeTable st')
->set('st.position','st.position + 1')
->leftJoin('st.SomeOtherTable sot ON st.some_id = sot.id')
->where('st.id <> ?', $someId)
->andWhere('sot.some_column = ?', $someValue)

The generated SQL for this is :
UPDATE some_table
LEFT JOIN some_other_table sot ON st.some_id = sot.id
SET position = position + 1, updated_at = 2011-04-13 11:01:03, updated_at = 2011-04-13 11:01:03
WHERE (id <> 4 AND some_column = 7)

Clearly here "updated_at" and "id" are ambiguous columns. Why the tables are not automatically aliased with unique aliases like in a SELECT statement, and the aliases written before the column name ?

Thanks.

While trying to load ~25M data fixtures (one big table with relations to 4 smaller ones, sfGuard included in that, row size around 500 bytes) in Symfony i ended up waiting around 80 minutes, while waiting i looked at what could have make it so dreadfully slow. Turns out, when Doctrine_Data_Load gets UnitOfWorks it executes save() on every new record. Save makes it's own transaction - not a problem if it's nested, but when this is the main transactions, 70000 of them make quite a difference. Remember - one of the main factors of DBMS speed is transactions/second.

I patched Doctrine_Data_Import to wrap everything in one transaction, and the results were great - from 80 minutes i got down to around 10. Still, not as fast as it should be but now it's usable. The time difference is notable also in smaller dumps.

Patch to speed up loading times included, would be great if you add it to trunk.

Please note - i have not checked this patch with any other setup or DBMS, please do so.

Also i have noticed something that might be a problem in much larger loads - if wrapping in a single transaction, my total memory usage went up for about 500M higher than in 70000 transactions. At some point, about 5 minutes in the process some kind of garbage collector fired and freed around 1 gig, so perhaps on larger dumps it might be a good idea to wrap the import not in one, but more transactions (like one transaction every 10000 operations).

I have used I18n behavior for my application using the following:

public function setTableDefinition()
{
        $this->setTableName('products');

        $this->hasColumn('id', 'integer', 4, 
            array('fixed' => true, 
                  'primary' => true, 
                  'autoincrement' => true));

        $this->hasColumn('permalink', 'string', 255,
            array('notnull' => true));
        
        $this->hasColumn('title', 'string', 255, 
            array('notnull' => true));
            
        $this->hasColumn('teaser', 'string', 255, 
            array('notnull' => true));
            
        $this->hasColumn('content', 'clob', 32767);
}

public function setUp()
{   
        $this->actAs('I18n', array(
                'fields' => array('title', 'teaser', 'content')
            )
        );
}

Doctrine has created two tables db named products and products_translation.

Insert and update of the record is working fine but when i perform a deletion of a record, the record is deleted from the products table but the translations stored in products translation table are not deleted.

View abstraction model

Hi, i've been using doctrine from about six months, i'm not an expert but i know the basics and this has been enough for me and my web-app requirements. The problem begins cause i need a kind of "dynamic table model" in other words an specific one table's abstraction, i thought implement a view for this purpose, but i can't figure out how define the BaseModel for the view to use it like a table, thus allowing the use of methods like save(), find() and build (logicals) relationships with others entities. in few words: can i build a table model from a query/view?, it is possible? i read the posts from above but this issue still being not realy clear at all for me.

me realy will apreciate any help, thanks in advance.

Regards.

In order to generate SQL from migrations, an event listener was attached to the migration system to monitor for preQuery and preExec events.

In an attempt to prevent the migration from additionally writing the query to the database, the skipOperation method was triggered, and supposedly allowed for n the execute() and exec() methods of Doctrine_Connection

 
$this->getAttribute(Doctrine_Core::ATTR_LISTENER)->preQuery($event);

if ( ! $event->skipOperation) {
    $stmt = $this->dbh->query($query);
    $this->_count++;
}
$this->getAttribute(Doctrine_Core::ATTR_LISTENER)->postQuery($event);

unfortunately setting this option in the event listener breaks execution of the migration system as the $count/$stmtn variables (used in the methods) are no longer defined, triggering E_NOTICE, and the fetch* methods (eg fetchColumn) also break as they are chaining methods without testing for the return. theerfore, even if the $stmnt variable was created as initially null, the system would still throw an E_FATAL as the NULL variable doest provide the ::fetchColumn() methods (etc).

Quite a serious flaw as 2 areas of code do not provide a consistent approach to how to work with events.

At present migrations only allow for direct manipulation of the underlying database. However, many enterprise release processes disallow automated manipulation of databases (especially Oracle) due to a number of reasons (eg placing different objects in different table spaces).

Because of this, it is preferable for developers to auto generate SQL and then hand over the specialist DBAs who may then filter/alter as needed on a per-environment basis.

this is currently very easy to achieve with initial database query generation, as outputting SQL is an option, but there is no such option for migration scripts. Therefore I would like to request this option to be added to the migration class.

Current export methods are inconsitent with index/constraint name suffix (defautl %_idx). Both createConstraintSql() and dropIndex() methods correctly set the suffix, but dropConstraint() and createIndexSql() do not.

this causes associated down() methods to fail when reverting changes to indexes/constraints

Erros occur in : Export.php - lines 137 and 473

Migration commands update the database without correcting the default tablename using pre-set tblename_format parameters in databases.yml.

There is a method for updating the tablename, but this appears to not be used by any script.

When using pessimistic locking as described in:

http://www.doctrine-project.org/projects/orm/1.2/docs/manual/component-overview:locking-manager:examples/zh

the locking manager locks the entire table rather than the specific object.

This should be clear from the attached patch which corrects the issue (assuming I have correctly interpreted the intention of pessimistic locking!).

The current behavior will have worked as expected for users but it will have locked far more than was intended and may thus have affected performance.

NB: I can confirm this works for non-composite keys but please review and test for composite keys as I have no such tables to hand.

Bug found when working on project ma-residence.fr. Data loading was repeated twice, and in the second run, empty rows were inserted in database, resulting in major headache in development team.

A same flush tree is built for each connections. It results in multiple loops of data load when there is more than one connection.

I.e.: generate_models_options and classPrefix option isn't forwared from CLI configuration to create table task (which uses the Manager's options).

Configuration:

Doctrine_Manager::getInstance()->setAttribute(Doctrine_Core::ATTR_MODEL_CLASS_PREFIX, 'Foo_');

Schema:

Bar:
columns:
baz: int

Fixtures:

Bar:
Bar_1:
baz: 1

Error on importing data: "Couldn't find class Bar."

Doctrine should use Foo_Bar class for Bar model instead Bar class.

Updating/saving an object fails when trying to update the nth-level of an existing object/relation where n is more than one level away from the root node of one-to-one relation chain. For example, with the yaml model below the following does not update the name of the Company object:

Image->Product->Category->Company->name = "Acme".

If "Product" has no changes it appears to stop checking for changes there. If we do a save on a one-to-many relation chain such as the following, it works fine:

Company->Category->Product->Image->name = "image1.jpg"

I was able to fix this issue by modifying the saveRelatedLocalKeys() function in UnitOfWork.php to use isModified() with deep=true:

if ($obj instanceof Doctrine_Record && $obj->isModified(true)) {

This works but I'm not sure if changing this has any repercussions on more complex queries.

—
detect_relations: true
options:
collate: utf8_general_ci
charset: utf8
type: InnoDB

Company:
tableName: Company
columns:
id:
type: integer(4)
primary: true
notnull: true
autoincrement: true
name:
type: string(45)

Category:
tableName: Category
columns:
id:
type: integer(4)
primary: true
notnull: true
autoincrement: true
companyId:
type: integer(4)
notnull: true
name:
type: string(45)
relations:
Company:
class: Company
local: companyId
foreign: id
foreignAlias: categories
onDelete: cascade
onUpdate: cascade
indexes:
fk_Category_Company1:
fields: [companyId]

Product:
tableName: Product
columns:
id:
type: integer(4)
primary: true
notnull: true
autoincrement: true
categoryId:
type: integer(4)
notnull: true
name:
type: string(45)
relations:
Category:
class: Category
local: categoryId
foreign: id
foreignAlias: products
onDelete: cascade
onUpdate: cascade
indexes:
fk_Product_Category1:
fields: [categoryId]

Image:
tableName: Image
columns:
id:
type: integer(4)
primary: true
notnull: true
autoincrement: true
productId:
type: integer(4)
notnull: true
name:
type: string(45)
relations:
Product:
class: Product
local: productId
foreign: id
foreignAlias: images
onDelete: cascade
onUpdate: cascade
indexes:
fk_Image_Product1:
fields: [productId]

see http://trac.symfony-project.org/ticket/4522

When using a behaviour with generateFile=true, som eproblematic issues occur:

Let's take for example i18n:

schema.yml
...
actAs:
fields: [text]
generateFiles: true
generatePath: <?php echo sfConfig::get('sf_lib_dir') ?>/model/doctrine/translations

1) The translation model class and it's base class are not created with generate.php (aka symfony doctrine:build --model), but every time a translation model is used. This is not the expected behaviour, because
a) this makes autoloading these classes impossible.
b) the APC cache always sees a new change time, and recaches the fiels. This results in a quick fragmentation of the cache with segmentation faults in the long run.
2) The permissions for the created files are wrong. Normal executable php scripts should not be writable by the web server (admitting that with the current implementation, this is not problem, because the files are created again for each request)
3) The path is hard coded. This brakes deployment by svn update, and I would prefer not to build models on a production server.

Proposal:

• generate the files at build model time
• hopefully find a solution with the path

I realize that most development resources are now in the new doctrine, but this issue (especially the apc fragmentation) is a huge problem for me. If you won't fix it, let me know, then I would try to propose a patch. I checked the code, and the building and behaviour internal part of doctrine are not too well documented, and my patch would be far from perfect.

Statements that return no result are not correctly closed in Doctrine_Hydrator_Graph::hydrateResultSet().

Oracle has limited number of opened cursors, and this bug prevents unsing doctrine in batch task like indexing models with sfSolrPlugin.

Oracle throws an error `ORA-01000: maximum open cursors exceeded : ` in my case after indexing only 100 records.

I'll send a pull request via github for this issue.

Tree hierarchy hydrators (Doctrine_Collection::toHierarchy and Doctrine_Array_Hierarchy_Driver::hydrateResultSet) are checking wheter the column `level` exists.

The level column can be aliased, and for oracle, it is required to do so. Therefor it is better to check, whether the aliased field level exists.

Patch included in pull request

I am using PHP 5.3.2 and PostgreSQL 8.4.5, Diem passed all checks in green - OK.

I started with "A week of Diem Ipsum" and all went ok until I reached building of blog engine. Blog engine example fails in step:

php symfony doctrine:migrate

with error message:

The following errors occurred:

    * SQLSTATE[42830]: Invalid foreign key: 7 ERROR: there is no unique constraint matching given keys for referenced table "article_translation". Failing Query: "ALTER TABLE article_translation_version ADD CONSTRAINT article_translation_version_id_article_translation_id FOREIGN KEY (id) REFERENCES article_translation(id) ON UPDATE CASCADE ON DELETE CASCADE NOT DEFERRABLE INITIALLY IMMEDIATE"
    * SQLSTATE[25P02]: In failed sql transaction: 7 ERROR: current transaction is aborted, commands ignored until end of transaction block. Failing Query: "CREATE INDEX article_image ON article (image)"
    * SQLSTATE[25P02]: In failed sql transaction: 7 ERROR: current transaction is aborted, commands ignored until end of transaction block. Failing Query: "CREATE INDEX article_author ON article (author)"
    * SQLSTATE[25P02]: In failed sql transaction: 7 ERROR: current transaction is aborted, commands ignored until end of transaction block. Failing Query: "CREATE INDEX article_translation_id ON article_translation (id)"
    * SQLSTATE[25P02]: In failed sql transaction: 7 ERROR: current transaction is aborted, commands ignored until end of transaction block. Failing Query: "CREATE INDEX article_translation_version_id ON article_translation_version (id)"

I removed i18n support in blog engine example and after that migrate went ok. But in Admin interface when I wanted to add
blog article, although Diem confirmed it saved article, article would not show up in the list, I checked db table, it was empty also.
Further, if I try to loremize, for 1 fixture I get no error but table is again empty, and for more fixtures then 1, Diem reports error:

SQLSTATE[25P02]: In failed sql transaction: 7 ERROR: current transaction is aborted, commands ignored until end of transaction block.

Again I reviewed the model and removed DmVersionable, migrated again and after that I could loremize or create articles without errors.

Additionally, not related directly to this blog engine example but doctrine related, I noticed errors in Diem Admin interface
itself when I try to access System->Configuration->Settings . If I access System settings over link
admin_dev.php/system/configuration/settings/index it shows settings. But when I click on
dmin_dev.php/system/configuration in menu path and after again to settings
admin_dev.php/system/configuration/settings/index, error is generated:

500 | Internal Server Error | Doctrine_Connection_Pgsql_Exception
SQLSTATE[08P01]: <>: 7 ERROR: bind message supplies 1 parameters, but prepared statement "pdo_stmt_00000008" requires 2

Are this bugs corrected?

They are some strange problems with hydration to array combined with aliases of columns and JOINS. Let's see this example:

$q = Doctrine_Query::create ()
->select('c.id AS id, c.path AS path, c.name AS name, cbc.product_count')
->from('Category c')
->leftJoin('c.CategoryBrowserCache cbc');

$categories = $q->fetchArray();

This example will throw exception: "The root class of the query (alias lc) must have at least one field selected."

OK. Let's change code a little bit. Let's add alias for cbc.product_count column too:

$q = Doctrine_Query::create ()
->select('c.id AS id, c.path AS path, c.name AS name, cbc.product_count AS product_count')
->from('Category c')
->leftJoin('c.CategoryBrowserCache cbc');

$categories = $q->fetchArray();
print_r($categories);

Now code executed without exception, BUT $q->fetchArray() returned only ONE (first) record hydrated to array. Other results were ignored.

Let's change code one more time:

$q = Doctrine_Query::create ()
->select('c.id, c.path, c.name, cbc.product_count AS product_count')
->from('Category c')
->leftJoin('c.CategoryBrowserCache cbc');
$categories = $q->fetchArray();
print_r($categories);

Like you see I just removed all aliases for columns for Category. Now code will be executed without exceptions, All results will be hydrated into array as expected to be. Actually the same result can be reached by removing at least one alias for any Category column.

Symfony Version 1.4.9 ORM: Doctrine

Schema.yml:

Table1:
actAs:
Timestampable: ~
options:
orderBy: sort_order ASC
columns:
name:

sort_order:

Table2:
columns:
table1_id:

value:

relations:
Table1:

This generates models and I can see the following: BaseTable?1.class.php: $this->option('sort_order', 'sort_order ASC');

BaseTable?2.class.php: No option for sort_order

But when I run the following, I get errors: Doctine::getTable('Table1')>createQuery('t')>leftJoin('t.Table2 t2').execute();

Error: Column not found: 1054 Unknown column 't2.sort_order' in 'order clause'

Looking at the sql executed, it included t1.sort_order ASC, but also incorrectly added t2.sort_order ASC as well even though it was never defined anywhere.

I am unsure if this is a Doctrine problem or a symfony one, so I will post i on both bug tracking systems.

I have a column that contains a serialized string of data. The accessor for that column unserializes the data and the mutator serializes. When calling copy the accessor is not used but the mutator is causing the mutator to fail because it receives a string instead of an object as its value.

The lines of code creating this problem follow:

 public function copy($deep = false)
 {
    $data = $this->_data; //does not use accessor
...
    $ret = $this->_table->create($data); // does use mutator
...
}

I have currently patched my copy function with the following:

// $data = $this->_data
$data = $this->toArray(false);

oci_free_statement should be changed to oci_cancel inside Doctrine_Adapter_Statement_Oracle::closeCursor(). Otherwise exception is thrown if HYDRATE_ON_DEMAND is used followed by foreach loop.

Doctrine2 should also be affected by this bug.

Change:
public function closeCursor()

To:
public function closeCursor()

Sorry for the lengthy explanation, couldn't make it more straight forward:

I have a model which is similiar to a nestet set but the tree structure needs to overlap:

For Model A, every Object A1 can have multiple descendant objects A2 and in turn can be a descendant of multiple objects A0.

Since I saw no way to do this with Nested-Set Relations (or Equal-Nested-Sets) I have set up my Model like this:

modules:
columns: ..<do not matter>
relations:
Children:
class: modules
refClass: modules_required
Parents:
class: modules
refClass: modules_required

modules_required:
columns: <do not matter here, just 2 foreign key columns>
relations:
Children:
Parents:

I needed to specify the Relations on both tables, to use onDelete/onUpdate CASCADE rules. Generated Models look fine, just as intended.
(Every Class has many Children and has many Parents...)

Now the strange part:
When I update an object of modules (say id=18), Doctrine issues the following queries:
DELETE FROM modules_required WHERE (required_id = ? AND module_id IN (?, ?, ?, ?, ?)) - (18, 25, 26, 32, 34, 35)
// where 25 to 35 are CHILDREN of 18
UPDATE modules_required SET required_id = ? WHERE module_id = ? AND required_id = ? - (25, 25, 10)
UPDATE modules_required SET required_id = ? WHERE module_id = ? AND required_id = ? - (26, 26, 10)
UPDATE modules_required SET required_id = ? WHERE module_id = ? AND required_id = ? - (32, 32, 10)
UPDATE modules_required SET required_id = ? WHERE module_id = ? AND required_id = ? - (34, 34, 10)
UPDATE modules_required SET required_id = ? WHERE module_id = ? AND required_id = ? - (35, 35, 10)
UPDATE modules_required SET required_id = ? WHERE module_id = ? AND required_id = ? - (25, 25, 12)
//where 10 and 12 are PARENTS of 18
and somewhen, Doctrine encounters an MySQL ERROR because of the previous update marathon:
SQLSTATE[23000]: Integrity constraint violation: 1062 Duplicate entry '25-25' for key 'PRIMARY'

The point is:
1. why is Doctrine trying to create self-referencing relations, and
2. why is it touching the relation at all, when i only did change some text fields in the object?

Is there a better way to solve my problem?

http://www.doctrine-project.org/projects/orm/1.2/docs/manual/dql-doctrine-query-language/en#group-by,-having-clauses

$q = Doctrine_Query::create()
->select('u.username')
->addSelect('COUNT(p.id) as num_phonenumbers')
->from('User u')
->leftJoin('u.Phonenumbers p')
->groupBy('u.id');

SELECT
u.id AS u__id,
u.username AS u__username,
COUNT(p.id) AS p__0
FROM user u
LEFT JOIN phonenumber p ON u.id = p.user_id
GROUP BY u.id

i should create

SELECT
COUNT(p.id) AS p__0
FROM user u
LEFT JOIN phonenumber p ON u.id = p.user_id
GROUP BY u.id

MSSQL doesnt support this use of group by sql. Id have to be in aggregrate function or group by. I do not need id but doctrine creates it in sql.

"invalid in the select list because it is not contained in either an aggregate function or the GROUP BY clause."

As soon as one enables the Searchable behavior like the following it is not possible anymore to create or update an entity.

public function setTableDefinition()

Unknown macro: { $this->setTableName('seartest_article'); $this->hasColumn('articleid', 'integer', 11, array( 'type' => 'integer', 'primary' => true, 'autoincrement' => true, 'unsigned' => true, 'notnull' => false, 'unique' => true, 'length' => 11 ) ); $this->hasColumn('title', 'string', 255, array( 'type' => 'string', 'notnull' => false, 'length' => 255 ) ); $this->hasColumn('description', 'string', null, array( 'type' => 'string', 'notnull' => false, 'length' => null ) ); $this->option('type', 'INNODB'); $this->option('collate', 'utf8_unicode_ci'); $this->option('charset', 'utf8'); }

public function setUp()

Unknown macro: { // Search support $search = new Doctrine_Template_Searchable(array( 'fields' => array('title', 'description'), 'batchUpdate' => false ) ); $this->actAs($search); $this->addListener(new SearchableTest_Model_ArticleListener()); }

After trying to create a new item the following error message appears (note the record class is called SearchableTest_Model_Article and the primary id column is named articleid):

Validation failed in class SearchableTest_Model_ArticleIndex 1 field had validation error: * 1 validator failed on articleid (unique)

Sample schema:

Blog:
actAs: [Versionable, Searchable]
columns:
id:

name: string
text: text

When dumping data of a schema with Versionable and/or Searchable Behaviour, the dump.yml will contain all data, including the *_version and *_index tables.

Trying to load the same .yml file results in Duplicate-Key constraint violations, as long as the data for the *_version and *_index tables is present.
The import is only successfull, when the data for these tables is discarded.
This leads to the issue, that one can only do a dump-load cycle, when the complete version history of a Model is discarded. Which makes dump-load for such models rather useless.

doing a
doctrine:build --all --and-load
with a schema.yml of

Foo:
tableName: foo
options:
charset: utf8
columns:
bar:
type: tinyint(1)
default: 0

results in a blank value generated for the default keyword, and the following error:

SQLSTATE[42000]: Syntax error or access violation: 1064 You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near ' PRIMARY KEY(id)) DEFAULT CHARACTER SET utf8 ENGINE = INNODB' at line 1. Failing Query: "CREATE TABLE foo (id BIGINT AUTO_INCREMENT, bar tinyint(1) DEFAULT , PRIMARY KEY(id)) DEFAULT CHARACTER SET utf8 ENGINE = INNODB". Failing Query: CREATE TABLE foo (id BIGINT AUTO_INCREMENT, bar tinyint(1) DEFAULT , PRIMARY KEY(id)) DEFAULT CHARACTER SET utf8 ENGINE = INNODB

I've also tried combinations of tinyint, tinyint(4), single-quoting the default value, and different default values.

Changing the type to int makes the issue disappear

I have continually run into a very particular bug when using OneToMany relationships between Doctrine tables. When attempting to call "link()" to generate a relationship between records in related tables, Doctrine attempts to set the ID of the "one" portion of the record to 0, then save it.

This is best demonstrated by the sample YML and PHP that I have attached. It establishes a OneToMany relationship where the "one" table has another foreign key constraint. This causes the DB to trigger a foreign key constraint error when Doctrine tries to set the ID to 0, making the error easier to see.

From looking into the relevant sections of the codebase, the following appears to be happening:

• Calling "link" properly adds the relationship to the "many" record.
• Calling "save" triggers the saving of the "many" record and all associated records.
• With the "one" record now listed as an associated record, its own "save" function is called as a result.
• Its "save" function works properly, except that for some reason, the ID key is not present in the $_data internal array (but oddly is present in the $_id array, allowing it to be returned in other places).
• Seeing a difference, Doctrine attempts to execute an "UPDATE" query setting the ID to 0.

I've made it this far in looking into it, but for the life of me I can't figure out what is triggering the identifier being reset in this case. I should note, however, that it happens consistently in every such situation on every server I've tested it on.

collaboration of vtamara@pasosdejesus.org and jeronimo0000@gmail.com

While we developed a tool with symfony 1.4 and postgresql database we found errors in the generated schema.yml which I describe below

1 - Error in generating of field size of varchars
2 - Error in the generation of date fields

We found the following solution

— Doctrine/Import/Pgsql.php.orig 2010-12-23 17:48:00.160271000 -0500
+++ Doctrine/Import/Pgsql.php 2010-12-23 18:01:59.252271002 -0500
@@ -168,11 +168,14 @@
$columns = array();
foreach ($result as $key => $val) {
$val = array_change_key_case($val, CASE_LOWER);

• if (strtolower($val['type']) === 'character varying') {
  + if (strtolower($val['type']) === 'varchar') { // get length from varchar definition $length = preg_replace('~.*\(([0-9]*)\).*~', '$1', $val['complete_type']); $val['length'] = $length; }

  + if ($val['type'] == 'date')

  { + $val['type'] = $val['complete_type'] = 'timestamp'; + }

$decl = $this->conn->dataDict->getPortableDeclaration($val);

This creates a new attribute constant Doctrine_Core::ATTR_USE_NATIVE_FLOAT and Doctrine_Core::ATTR_USE_NATIVE_DOUBLE. This will allow the setting of attributes of use_native_float = true and use_native_double = true. With these set to true in MySQL of the generated sql will no longer Make FLOAT(18,2) and will make it just FLOAT that is a true floating point the same thing with DOUBLE except it is now a true double precision floating point.

Proper adjustments are also made to MySQL, Oracle and Postgresql to use native floating point declarations to define both single and double precision floating point data types.

I have attached a patch to fix the floating point field types.

Example RawSQL:

$q = new Doctrine_RawSql();
    $q->select('{k.*}')
          ->from('SHP_MANDANT_KATEGORIE k')
          ->addComponent('k', 'ShpMandantKategorie k')
          ->where( 'k.id_mandant=' . $this->getIdMandant() )
          ->andWhere( 'k.id_parent=' . $this->getIdMandantkategorie() )
          ->andWhere( 'k.aktiv=1' )
          ->orderBy( 'k.sortorder' ); 

$q->count() generates:

SELECT COUNT(*) as num_results 
FROM (SELECT DISTINCT k.id_mandantkategorie 
              FROM SHP_MANDANT_KATEGORIE k 
              WHERE k.id_mandant=2 AND k.id_parent=1520 AND k.aktiv=1) as results

The illegal Part ist the "as results" at the end...

'Pending join conditions' are used by listeners to inject extra SQL conditions into a query. They are often used to add basic constraints on every query. An example is the bundled SoftDelete template. Its listener adds extra constraints such as s.deleted_at IS NULL to a query, to make sure that deleted rows are never retrieved on a query.

However, in the emitted SQL, Doctrine_Query does not use parentheses to group normal SQL conditions together. The pending join condition is simply added to the string without encapsulating existing expressions. This makes it possible to bypass the pending join conditions entirely by using the OR operator.

Example

For instance, the following query exhibits this problem:

$query = Doctrine_Query::create()
->from("SoftDeleteTest")
->where("name=?", "faulty")
->orWhere("name=?", "faulty");

This query emits the following SQL:

SELECT s.name AS s_name, s.deleted_at AS s_deleted_at FROM soft_delete_test s WHERE (s.name = 'faulty' OR s.name = 'faulty' AND (s.deleted_at IS NULL))

which returns also a deleted row.

Expected behavior

One would expect the pending join conditions always to hold, and to have precedence over regularly added SQL conditions. This could be accomplished in the most simple fashion by:

SELECT s.name AS s_name, s.deleted_at AS s_deleted_at FROM soft_delete_test s WHERE ( ( s.name = 'faulty' OR s.name = 'faulty' ) AND (s.deleted_at IS NULL));

As the existing expressions are now encapsulated by parentheses, it is no longer possible to bypass the pending join conditions injected by the query listener.

Full test case details:

init.sql

create database softdelete;
grant all privileges on softdelete.* to softdelete@localhost identified by 'uahwqeruwer';

use softdelete;
CREATE TABLE soft_delete_test (name VARCHAR(255), 
    deleted_at DATETIME DEFAULT NULL, 
    PRIMARY KEY(name)) ENGINE = INNODB;

insert into soft_delete_test values ('fine', null);
insert into soft_delete_test values ('faulty', now());

run.php

<?php

require "./1.2.3/lib/Doctrine.php";

spl_autoload_register(array('Doctrine', 'autoload'));

require "SoftDeleteTest.php";

$conn = Doctrine_Manager::connection("mysql://softdelete:uahwqeruwer@localhost/softdelete");
$conn->setAttribute(Doctrine::ATTR_USE_DQL_CALLBACKS, true);

$query = Doctrine_Query::create()
    ->from("SoftDeleteTest")
    ->where("name=?", "faulty")
    ->orWhere("name=?", "faulty");

$found = $query->execute();
foreach ($found as $f) {
    echo "ERROR! Found a deleted row: $f->name\n";
}
echo "Done.\n";

SoftDeleteTest.php (copied from Doctrine manual)

<?php

class SoftDeleteTest extends Doctrine_Record
{
    public function setTableDefinition()
    {
        $this->hasColumn('name', 'string', null, array(
                'primary' => true
            )
        );
    }

    public function setUp()
    {
        $this->actAs('SoftDelete');
    }
}

If I do toArray(true) on record with realtions and later fromArray($array, true) on with same data unnessesary calls to database are made.

$message = new Message();
$message->Sender = new User(); // if i leave out this line sender will first get loaded from database and then overwritten with provided data
$message->Receiver = new User();   // if i leave out this line receiver will first get loaded from database and then overwritten with provided data
$message->fromArray($data);

In Doctrine_Record::fromArray()

if ($deep && $this->getTable()->hasRelation($key)) {
    if ( ! $this->$key) {                                           --> data gets loaded from db here, refreshRelated is not even executed.
        $this->refreshRelated($key);
    }
...
}

Is this desired behavour? Wouldnt it be smarter to create empty object automaticly instead of loading it from db?
Also have look at http://trac.doctrine-project.org/ticket/1434

I'm using doctrine and some of mysql's spatial functions. I need to specify spatial index for my tables.
My declaration looks like:

Geometry:
tableName: geometry
options:
type: MyISAM
collate: utf8_unicode_ci
charset: utf8
columns:
id:
type: integer(4)
primary: true
autoincrement: true
geometry:
type: geometry
notnull: true
indexes:
geometry:
fields:

• geometry
  type: spatial

Exporting this definitions throws an exception: Unknown type spatial for index geometry_idx

Hi,

My Doctrine 1.2 is integrated inside CodeIgniter as a hook and I know that my char-set is utf8 with collation utf8_unicode_ci.

I have two YAML files, one for creating the DB and its tables and one to load some test data. My data can contain French accents (çéïë...). In my schama.yml I have correctly specified the collation and char-set:

options:
type: INNODB
charset: utf8
collate: utf8_unicode_ci

I double checked the settings in phpMyAdmin, everything is correct.

When I run my doctrine script from commandline to load my fixture to populate one of tables, all the French accents are replaced by junk!

Am I missing a setting or configuration or is there a bug in Doctrine?

I appreciate any help. Cheers.

P.S. Everything else works like a charm

Patch for Doctrine ..... to identify in some cases autoincremented
fields.
[Solution found thanks to Vladimir Tamara - vtamara AT pasosdejesus DOT org]

Doctrine/Import/Oracle.php

// Heuristic to check autoincremented fields.
// We check if there is a trigger on the field.
// We could also check if there is a sequence on the field.
// Side effect: slower generation of scheme
$q = "SELECT * FROM all_trigger_cols WHERE
table_name='$table' AND column_name='" . $val['column_name'];

$res2 = $this->conn->fetchColumn($q);
if (count($res2) > 0)

}

return $descr;

File Doctrine/Data/Import.php, line #80
if (end($e) == 'yml')

So, if the file is a .json (for exemple), it will be impossible to load it, even if we have specified "json" as format parameter.
And it sucks.

The fix would just be to change the line to :
if (end($e) == $this->getFormat())

When you call a stored procedure from a schema other than that of the current connection: [schema_name].[stored_procedure_name]([stored_procedure_arguments]) doctrine miss understands the string and throws a "Couldn't get short alias for" exception.

I fixed this by adding some more regex to the getExpressionOwner method of the Query Class.

I will post the patch shortly.

Will Ferrer

With a valid Json file :
PHP Warning: explode() expects parameter 2 to be string, object given in Doctrine/Doctrine/Import/Schema.php on line 381
PHP Catchable fatal error: Object of class stdClass could not be converted to string in Doctrine/Doctrine/Import/Schema.php on line 391

It's due to this line, line, in Doctrine/Parser/Json.php (#65) :
$json = json_decode($contents);

It should be:
$json = json_decode($contents, true);

Because casting the result as array will only affect the top-level element. You must use the second parameter of json_decode() to force every objects (including sub-objects) to be converted to indexed arrays.

Doctrine_Task_BuildAllReload never calls generate models-from-yaml. This does not coincide with the logic of Doctrine_Task_BuildAll and Doctrine_Task_BuildAllLoad.

BuildAllReload suggests that it will be building all (everything) and then reloading the database.
But instead it only rebuilds the database and loads data

public function __construct($dispatcher = null)
    {
        parent::__construct($dispatcher);

        $this->rebuildDb = new Doctrine_Task_RebuildDb($this->dispatcher);
        $this->loadData = new Doctrine_Task_LoadData($this->dispatcher);
        
        $this->requiredArguments = array_merge($this->requiredArguments, $this->rebuildDb->requiredArguments, $this->loadData->requiredArguments);
        $this->optionalArguments = array_merge($this->optionalArguments, $this->rebuildDb->optionalArguments, $this->loadData->optionalArguments);
    }
    
    public function execute()
    {
        $this->rebuildDb->setArguments($this->getArguments());
        $this->rebuildDb->execute();
        
        $this->loadData->setArguments($this->getArguments());
        $this->loadData->execute();
    }

Instead, I think it would be more efficient and understanding to follow the same logic as build-all and build-all-load by calling drop-db and build-all-load.

public function __construct($dispatcher = null)
    {
        parent::__construct($dispatcher);

        $this->dropDb = new Doctrine_Task_DropDb($this->dispatcher);
        
        $this->buildAllLoad = new Doctrine_Task_BuildAllLoad($this->dispatcher);
        
        $this->requiredArguments = array_merge($this->requiredArguments, $this->dropDb->requiredArguments, $this->buildAllLoad->requiredArguments);
        $this->optionalArguments = array_merge($this->optionalArguments, $this->dropDb->optionalArguments, $this->buildAllLoad->optionalArguments);
    }
    
    public function execute()
    {
        $this->dropDb->setArguments($this->getArguments());
        $this->dropDb->execute();
        
        $this->buildAllLoad->setArguments($this->getArguments());
        $this->buildAllLoad->execute();
    }

I attached a patch with the above changes... I got a little lost in the test area for Doctrine_CLI, so that is not included = )