This PR was merged into the 3.4 branch.
Discussion
----------
[DI] Fix non-string class handling in PhpDumper
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #25353
| License | MIT
| Doc PR | -
Commits
-------
28f00866b1 Ensure that inlined services with parameterized class name can be dumped
730b156f35 [DI] Fix non-string class handling in PhpDumper
This PR was merged into the 3.4 branch.
Discussion
----------
[DI] Force root-namespace for function calls in the dumper container
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Was not needed before 3.4, but now that the dumped container is namespaced, it is required.
Commits
-------
f87380c22a [DI] Force root-namespace for function calls in the dumper container
This PR was merged into the 3.3 branch.
Discussion
----------
[Serializer] Unset attributes when creating child context
| Q | A
| ------------- | ---
| Branch? | 3.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
In some cases, the `attributes` key isn't overrode when creating the context passed to nested normalizers.
It's definitely a bug, but an attacker cannot access to non public data (ignored attributes are checked before the `attributes` key). However some data that must be public may be missing as highlighted by the test.
I've introduced the initial bug here: https://github.com/symfony/symfony/pull/18834
Commits
-------
4ff9d99f23 [Serializer] Unset attributes when creating child context
This PR was merged into the 3.4 branch.
Discussion
----------
[Yaml] do not evaluate PHP constant names
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #25307
| License | MIT
| Doc PR |
PHP constant identifiers must be strings anyway. Thus, we only need to
parse quoted strings, but do not have to evaluate the data types.
Commits
-------
956287be72 do not evaluate PHP constant names
This PR was merged into the 3.4 branch.
Discussion
----------
[Serializer] improved CsvEncoder::decode performance
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| License | MIT
Improved CsvEncoder::decode performance by caching duplicate count calls.
Blackfire profiles before and after change tested on collection of 10000 elements:
[before] https://blackfire.io/profiles/9c08f789-cd29-4eae-92c8-046e3849a2b8/graph
[after] https://blackfire.io/profiles/a17bfb6b-ef82-41ee-9edd-9403f829d6ab/graph
Commits
-------
3b910a9fad [Serializer] improved CsvEncoder::decode performance by caching duplicate count calls
This PR was squashed before being merged into the 4.1-dev branch (closes#24375).
Discussion
----------
[Serializer] Serialize and deserialize from abstract classes
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | ø
| License | MIT
| Doc PR | Not yet
This PR adds a feature in the Serializer: allow to serialize and de-serialize abstract classes. Such feature is especially useful when dealing with domain objects.
# Example
Let's take the example of the following objects:
- `CodeRepository` defines a set of properties like `name` and `url`
- `GitHubCodeRepository` and `BitBucketCodeRepository` extends from the abstract `CodeRepository` class and adds a few properties.
- `Project` has a relation with a `codeRepository`, which has a type `CodeRepository`.
At the moment, the serializer can't serialize/deserialize correctly this `Project` object has it doesn't know how to deal with this `CodeRepository` abstract object.
This feature allows the serializer to deal with such situation. The `ObjectNormalizer` has now access to a `ClassDiscriminatorResolver` that knows, for a given abstract class:
- Is the "type" property it needs to read/write to uniquely identify each sub-class
- What's the name of the "type" for each sub-class mapping
# Usage without Framework Bundle
```php
$discriminatorResolver = new ClassDiscriminatorResolver();
$discriminatorResolver->addClassMapping(CodeRepository::class, new ClassDiscriminatorMapping('type', [
'github' => GitHubCodeRepository::class,
'bitbucket' => BitBucketCodeRepository::class,
]));
$serializer = new Serializer(array(new ObjectNormalizer(null, null, null, null, $discriminatorResolver)), array('json' => new JsonEncoder()));
$serialized = $serializer->serialize(new GitHubCodeRepository());
// {"type": "github"}
$repository = $serializer->unserialize($serialized, CodeRepository::class, 'json');
// GitHubCodeRepository
```
# Usage with the Framework Bundle
```yaml
framework:
serializer:
discriminator_class_mapping:
App\CodeRepository:
type_property: type
mapping:
github: App\GitHubCodeRepository
bitbucket: App\BitBucketCodeRepository
```
# Usage with Annotations/XML/YAML
```php
use Symfony\Component\Serializer\Annotation\DiscriminatorMap;
/**
* @DiscriminatorMap(typeProperty="type", mapping={
* "first"="Symfony\Component\Serializer\Tests\Fixtures\AbstractDummyFirstChild",
* "second"="Symfony\Component\Serializer\Tests\Fixtures\AbstractDummySecondChild"
* })
*/
abstract class AbstractDummy
{
public $foo;
public function __construct($foo = null)
{
$this->foo = $foo;
}
}
```
# TODO
- [x] Working as standalone
- [x] Working with the framework bundle
- [x] Tests on mapping classes
Commits
-------
4c6e05b7ee [Serializer] Serialize and deserialize from abstract classes
This PR was merged into the 4.1-dev branch.
Discussion
----------
[DoctrineBridge] DoctrineDataCollector comments the non runnable part of the query
| Q | A
| ------------- | ---
| Branch? | 4.1
| Bug fix? | yes
| New feature? | no <!-- don't forget to update src/**/CHANGELOG.md files -->
| BC breaks? | no
| Deprecations? | no <!-- don't forget to update UPGRADE-*.md files -->
| Tests pass? | yes
| Fixed tickets | #24782
| License | MIT
| Doc PR | symfony/symfony-docs#... <!--highly recommended for new features-->
![img_2932](https://user-images.githubusercontent.com/3451634/33648180-f6c7a5ac-da58-11e7-8bf8-95fc943d16ff.jpeg)
I think the idea in this ticket is good and it should be implemented. Could we go further by adding more things to this feature, or will it be ok to just comment out the un-needed part to make the kiri
![kiri](https://user-images.githubusercontent.com/3451634/33648278-5eccc830-da59-11e7-8034-a1b9efee7673.png)
(french joke for query) runnable ?
Commits
-------
42760d0d7f [DoctrineBridge] DoctrineDataCollector comments the non runnable part of the query
This PR was merged into the 3.4 branch.
Discussion
----------
[FrameworkBundle][Cache] register system cache clearer only if it's used
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #25326
| License | MIT
| Doc PR |
Commits
-------
093eb3d40d register system cache clearer only if it's used
This PR was merged into the 4.1-dev branch.
Discussion
----------
[Process] Create a "isTtySupported" static method
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | _none_
| License | MIT
| Doc PR | _none_
Currently, there is no way to enable the TTY mode without risking an exception. This PR extracts the code checking for TTY support and provides it in a `isTtySupported` static method.
Now we can enable the TTY mode everywhere it's available without risking an exception:
```php
$process = (new Process)->setTty(Process::isTtySupported());
```
_Old comment_:
> I'm targeting the 2.7 branch since this is not really a new feature, just a little refactoring of the existent code, and it is fully backward compatible.
Commits
-------
a1398f6de4 Create a "isTtySupported" static method
This PR was merged into the 2.7 branch.
Discussion
----------
[ExpressionLanguage] throw an SyntaxError instead of an undefined index notice
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no <!-- don't forget to update src/**/CHANGELOG.md files -->
| BC breaks? | no
| Deprecations? | no <!-- don't forget to update UPGRADE-*.md files -->
| Tests pass? | yes
| Fixed tickets | #25214
| License | MIT
| Doc PR | none
I think this is a bug when the components throws a notice instead of an exception.
it's too early and too dark to see something outside so here is my couch :
![img_2915-2](https://user-images.githubusercontent.com/3451634/33592448-6b514050-d98b-11e7-8086-bc6e6b6e6e82.jpg)
Commits
-------
78abc89648 [ExpressionLanguage] throw an SyntaxError instead of letting a undefined index notice
This PR was merged into the 3.3 branch.
Discussion
----------
[DependencyInjection] Prevent a loop in aliases within the `findDefinition` method
| Q | A
| ------------- | ---
| Branch? | 3.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #25338
| License | MIT
| Doc PR | ø
This prevents an infinite loop going when aliases reference themselves. This is based on 3.3 as the "normalized ID" changed to allow non-lowercase names. Fixing this in 2.7 would mean a merge conflict that IMO is not worth it.
Commits
-------
22f35239a4 Prevent a loop in aliases within the `findDefinition` method
This PR was merged into the 3.4 branch.
Discussion
----------
[DX][HttpKernel] Throw a sensible exception when controller has been removed
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #25335
| License | MIT
| Doc PR | ø
Following on #25201, we need to throw the same kind of sensible exception when the controller service is not found.
Commits
-------
458d63fbb9 Throw a sensible exception when controller has been removed
This PR was merged into the 4.1-dev branch.
Discussion
----------
[Workflow] Introduce a Workflow interface
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | yes
| Fixed tickets | #23910
| License | MIT
| Doc PR | todo
@chalasr I think all the points you made in 23910 has been done. Needs to update the docs too.
Commits
-------
e8351d8 [Workflow] Introduce a Workflow interface
This PR was merged into the 3.4 branch.
Discussion
----------
Remove Exclusive Lock That Breaks NFS Caching
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #25336
| License | MIT
| Doc PR | n/a
#24960 introduced an issue with NFS mounts that do not support exclusive locks. This reverts that change.
FYI @kalessil
Commits
-------
a7ac100 Remove LOCK_EX That Breaks Cache Usage on NFS
This removes the exclusive lock that was introduced in #24960.
NFS File Systems do not support exclusive locking, and generates a lot
of errors every time you try to do anything with che cache.