This PR was squashed before being merged into the 3.3-dev branch (closes#21937).
Discussion
----------
[DependencyInjection] Handle void return types in closure-proxy
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | N/A
| License | MIT
| Doc PR | N/A
I recently got an error when registering an event listener that specifies a `void` return type. Dumping the container generates a closure proxy that always returns a value, which then conflicts with the return type hint.
E.G the following code is generated (some class names removed for readability)
```
$instance->addListener('kernel.view', /** @closure-proxy ... */ function (...\GetResponseForControllerResultEvent $event): void {
return ${($_ = isset($this->services[listener']) ? $this->services['listener'] : $this->get('listener')) && false ?: '_'}->onKernelView($event);
}, 128);
```
This then causes the error `A void function must not return a value in ...`
So void return types should be handled by removing the `return` inside the closure
Commits
-------
a5c5ad1 [DependencyInjection] Handle void return types in closure-proxy
This PR was squashed before being merged into the 3.3-dev branch (closes#21763).
Discussion
----------
[DI] Replace wildcard-based methods autowiring by `@required` annotation
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no (affects things that are only on master)
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
While playing a bit with new features in master around DI configuration, several people around me got bitten by wildcard-based autowiring. The typical example is adding `autowire: [set*]` in `_defaults`: use that on `resource: ../src/Command/` PSR4-based loading and boom, `setApplication` and `setHelperSet` will now be wrongly called. You could tell me "of course, don't to that" - but being bitten so early on a master-only feature makes me really unconfident that this will be easy enough for people after the release.
If wildcard-based autowiring is removed, then I don't see anymore the need for allowing arrays as in `autowire: [setFoo,getBar]`. Moreover, this array syntax has a core DX issue: it's a dead end as far as the learning curve is concerned. You learn it, then when becoming a more advanced dev, someone teaches you that you'd better use another syntax: explicit wiring.
And in fact, we don't need it at all, because something else already exists: just declare a method call, but don't define its arguments. If `autowire: true` is set, then the AutowiringPass already fills in the holes. There is only one tweak required to make this work: don't autowire optional arguments for method calls - or that'd be a BC break. To my PoV that's even better: this makes autowiring fit a "do the minimum to make it work" strategy. A really good one to me.
But there is still an issue: wildcard-based autowiring fits a need. Namely, it allows one to define a convention (eg. `'set*'`), and have all such methods that follow the convention be autowired. To me, this looks like doing it reverse (the DI config should adapt to the code, not reverse). So, to fill this need, let the declaration be in the source: just use an annotation!
This PR adds support for the `@required` annotation, borrowed from the Spring framework:
https://www.tutorialspoint.com/spring/spring_required_annotation.htm
Using the annotation is totally optional of course. If you do, *and if autowiring is on*, then it'll be autowired. If you don't, nothing changes: do manual wiring.
Even when not using autowiring, the annotation is still a nice hint for the consumer of your classes: it tells the reader that this method needs to be called for correct instantiation - thus lowering one drawback of setter injection (discoverability).
The implementation of the annotation parsing is done using a few regexp (no dep on any complex parser) - and works with inheritance, by leveraging the `@inheritdoc` tag (the default behavior being to *not* inherit anything from parent methods).
All in all, looking at the diff stats, it makes everything simpler. Good sign, isn't it?
Commits
-------
f286fcc25f [DI] Replace wildcard-based methods autowiring by `@required` annotation
9081699980 Revert "minor #21315 [DI][FrameworkBundle] Show autowired methods in descriptors (ogizanagi)"
This PR was merged into the 3.3-dev branch.
Discussion
----------
Fix DI test
| Q | A
| ------------- | ---
| Branch? | master
| Tests pass? | yes
Should fix appveyor/travis builds
Commits
-------
8740e44086 Fix DI test
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DependencyInjection] make the service container builder register its own self referencing definition
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | ~
| License | MIT
| Doc PR | ~
Commits
-------
9c97496b5f [DependencyInjection] make the service container builder register the definition of its related service container service (and aliases) in order to make compiler passes be able to reference the special service_container service.
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DI] Always consider abstract getters as autowiring candidates
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes (a missing part of getter autowiring really)
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
When a definition is set to be autowired with no method explicitly configured, we already wire the constructor.
We should also autowire abstract getters - with the same reasoning that makes us autowire the constructor: without concrete getters, the class is unusable. This just makes it usable again.
Commits
-------
8f246bde1d [DI] Always consider abstract getters as autowiring candidates
* 3.2:
Permit empty suffix on Windows
fixed CS
[FrameworkBundle] Remove unused import
[Console][Table] fixed render when using multiple rowspans.
add docblocks for Twig url and path function to improve ide completion
check for circular refs caused by method calls
[Serializer] fix upper camel case conversion (see #21399)
[DI] Auto register extension configuration classes as a resource
[Console] Updated phpdoc on return types
* 2.8:
Permit empty suffix on Windows
[Console][Table] fixed render when using multiple rowspans.
add docblocks for Twig url and path function to improve ide completion
check for circular refs caused by method calls
[Serializer] fix upper camel case conversion (see #21399)
[DI] Auto register extension configuration classes as a resource
[Console] Updated phpdoc on return types
* 2.7:
Permit empty suffix on Windows
[Console][Table] fixed render when using multiple rowspans.
add docblocks for Twig url and path function to improve ide completion
check for circular refs caused by method calls
[Serializer] fix upper camel case conversion (see #21399)
[DI] Auto register extension configuration classes as a resource
[Console] Updated phpdoc on return types
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DI] Deprecate underscore-services in YamlFileLoader
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | no
| Fixed tickets | -
| License | MIT
| Doc PR | -
As discussed when introducing `_defaults`
Commits
-------
7781082587 [DI] Deprecate underscore-services in YamlFileLoader
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DependencyInjection] Add support for named arguments
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | todo
This PR introduces named arguments for services definitions. It's especially useful to inject parameters in an autowired service. It is (at least partially) an alternative to #21376 and #20738.
Usage:
```yml
services:
_defaults: { autowire: true }
Acme\NewsletterManager: { $apiKey: "%mandrill_api_key%" }
# Alternative (traditional) syntax
services:
newsletter_manager:
class: Acme\NewsletterManager
arguments:
$apiKey: "%mandrill_api_key%"
autowire: true
```
```php
use Doctrine\ORM\EntityManager;
use Psr\Log\LoggerInterface;
namespace Acme;
class NewsletterManager
{
private $logger;
private $em;
private $apiKey;
public function __construct(LoggerInterface $logger, EntityManager $em, $apiKey)
{
$this->logger = $logger;
$this->em = $em;
$this->apiKey = $apiKey;
}
}
```
Commits
-------
8a126c8537 [DI] Deprecate string keys in arguments
2ce36a6074 [DependencyInjection] Add a new pass to check arguments validity
6e501296f9 [DependencyInjection] Add support for named arguments
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DI] Add prototype services for PSR4-based discovery and registration
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | to be done
Talking with @dunglas, we wondered if this could be a good idea, as a more general approach to folder-based service registration as done in [DunglasActionBundle](https://github.com/dunglas/DunglasActionBundle/blob/master/DependencyInjection/DunglasActionExtension.php).
So here is the implementation.
This allows one to define a set of services as such:
```yaml
services:
App\:
resources: ../src/{Controller,Command} # relative to the current file as usual
autowire: true # or any other attributes really
```
This looks for php files in the "src" folder, derivates PSR-4 class names from them, and uses `class_exists` for final discovery. The resulting services are named after the classes found this way.
The "resource" attribute can be a glob to select only a subset of the classes/files.
This approach has several advantages over [DunglasActionBundle](https://github.com/dunglas/DunglasActionBundle/blob/master/DependencyInjection/DunglasActionExtension.php):
- it is resilient to missing parent classes (see test case)
- it loads classes using the normal code path (the autoloader), thus play well with them (e.g. if one registered a special autoloader).
- it is more predictable, because it uses discovered paths as the only source of ids/classes to register - vs relying on `get_declared_classes`, which would make things context sensitive.
Fits well with current initiatives to me.
Commits
-------
03470b788e [DI] Add "psr4" service attribute for PSR4-based discovery and registration
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DI] Deprecate (un)setting pre-defined services
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | no
| Fixed tickets | #19192
| License | MIT
| Doc PR | -
This PR is the subset of #19668 that fixes#19192: it deprecates (un)setting pre-defined services.
This opens the path to some optimizations in the dumped container in 4.0.
Commits
-------
fdb2140b81 [DI] Deprecate (un)setting pre-defined services
This PR was squashed before being merged into the 3.3-dev branch (closes#21194).
Discussion
----------
[Yaml] Add tags support
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | yes
| Fixed tickets | https://github.com/symfony/symfony/issues/21185
| License | MIT
| Doc PR |
This PR adds custom tags support to the Yaml component.
Symfony tags (`!!binary`, `!str`, etc.) are still managed in the parser to have a lighter diff but we'll be able to convert them later if we want to.
The primary addition of this PR is the `TagInterface`:
```php
interface TagInterface
{
public function construct(mixed $value): mixed;
}
```
It can be used to register custom tags. An example that could be used to convert [the syntax `=iterator`](https://github.com/symfony/symfony/pull/20907#issuecomment-270728216) to a tag:
```php
final class IteratorTag implements TagInterface
{
public function construct(mixed $value): mixed
{
return new IteratorArgument($value);
}
}
$parser = new Parser(['iterator' => new IteratorTag()]);
```
If you think this is too complex, @nicolas-grekas [proposed an alternative](https://github.com/symfony/symfony/issues/21185#issuecomment-271074840) to my proposal externalizing this support by introducing a new class `TaggedValue`.
Commits
-------
4744107 [Yaml] Add tags support
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DI] Allow to count on lazy collection arguments
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | https://github.com/symfony/symfony/pull/21450#issuecomment-275931764
| License | MIT
| Doc PR | todo (with https://github.com/symfony/symfony-docs/issues/7336)
When using the new iterator feature of the DI component to lazy load collection, we always know the number of arguments in the collection (only the invalidBehavior set to `IGNORE_ON_INVALID_REFERENCE` may change this number). So we are able to generate and use a `RewindableGenerator` implementing `\Countable` by computing this value ahead.
So, in a service accepting `array|iterable`, like in the `GuardAuthenticationListener` (see #21450):
```php
class GuardAuthenticationListener implements ListenerInterface
{
private $guardAuthenticators;
/**
* @param iterable|GuardAuthenticatorInterface[] $guardAuthenticators The authenticators, with keys that match what's passed to GuardAuthenticationProvider
* @param LoggerInterface $logger A LoggerInterface instance
*/
public function __construct($guardAuthenticators, LoggerInterface $logger = null)
{
// ...
}
public function handle(GetResponseEvent $event)
{
if (null !== $this->logger) {
$context = array()
if (is_array($this->guardAuthenticators) || $this->guardAuthenticators instanceof \Countable) {
$context['authenticators'] = count($this->guardAuthenticators);
}
$this->logger->debug('Checking for guard authentication credentials.', $context);
}
// ...
}
}
```
we still keep the ability to call count without loosing the lazy load benefits.
Commits
-------
f23e460 [DI] Allow to count on lazy collection arguments
This PR was merged into the 3.3-dev branch.
Discussion
----------
[DI] Add getter injection
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20657
| License | MIT
| Doc PR | symfony/symfony-docs#7300
Getter overriding by the container will allow a new kind of dependency injection which enables easier laziness and more immutable classes, by not requiring any corresponding setter. See linked issue for more.
This is WIP:
- [x] wire the concept
- [x] dump anonymous classes with PhpDumper
- [x] generate at runtime in ContainerBuilder::createService
- [x] tests
- [x] make it work on PHP 5
Commits
-------
cb498580d1 [DI] Add getter injection
This PR was squashed before being merged into the 3.3-dev branch (closes#21348).
Discussion
----------
[DependencyInjection] Yaml: check if $tags is an array before using it
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
Commits
-------
1c9b5c9 [DependencyInjection] Yaml: check if is an array before using it
