This PR was squashed before being merged into the 4.1-dev branch (closes#25732).
Discussion
----------
[Console] Add option to automatically run suggested command if there is only 1 alternative
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
When mistyping a console command, you get an error giving suggested commands.
If there is only 1 alternative suggestion, this PR will give you the option to run that command instead. This makes it easier to run the correct command without having to re-type/copy-paste/update the previous run command
![console](https://user-images.githubusercontent.com/144858/34724377-4b46c726-f556-11e7-94a3-a9d7c9d75e74.gif)
Commits
-------
83d52f02f9 [Console] Add option to automatically run suggested command if there is only 1 alternative
This PR was squashed before being merged into the 4.1-dev branch (closes#26085).
Discussion
----------
Deprecate bundle:controller:action and service:method notation
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | yes
| Fixed tickets | #25910
| License | MIT
| Doc PR |
The `a::b` notation had some awkward limitations. It supported `MyControllerClass::method` where `MyControllerClass` is either plain class or a service with the same name but the class must exists. This meant it did NOT support `my_service_controller_id::method` because the `class_exists` check would fail at the wrong point in time. But it did support services where class name == id, i.e. the new auto registration based psr naming. This made it very confusing.
I enhanced the `a::b` notation to be very straight forward:
- if `a` exists as a service then use `a` as a service
- otherwise try to use `a` as a class, i.e. `new $a()`
- otherwise check if a::b is a static method (only relevant when the class is abstract or has private contructor). this was potentially supported when using array controller syntax. it now works the same when using the `::` string syntax, like in php itself. since it only happens when nothing else works, it does not have any performance impact.
The old `a:b` syntax is deprecated and just forwards to `a::b` now internally, just as bundle:controller:action.
In general I was able to refactor the logic quite a bit because it always goes through `instantiateController` now.
Spotting deprecated usages is very easy as all outdated routing configs will trigger a deprecation with the DelegatingLoader and it will be normalized in the dumped routes. So you don't get a deprecation again in the ControllerResolver. But if the controller does not come from routing, e.g. twigs render controller function, then it will still be triggered there.
- [x] deprecate `a🅱️c`
- [x] deprecate `a:b`
- [x] update existing references to `a::b`
- [x] fix tests
- [x] fix/add support for static controllers
- [x] add support for closures as controllers
- [x] update Symfony\Component\Routing\Loader\ObjectRouteLoader
- [x] deprecate \Symfony\Bundle\FrameworkBundle\Controller\ControllerNameParser but we still need to use it in several places for BC.
- [x] add changelog/upgrade
- [x] update controller.service_arguments logic for double colon controller syntax
Commits
-------
f8a609cdbd Deprecate bundle:controller:action and service:method notation
This PR was merged into the 4.1-dev branch.
Discussion
----------
[MonologBridge] Allow to change level format in ConsoleFormatter
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
I want to use this format `%datetime% <comment>[%level_name%]</comment> %start_tag%%message%%end_tag%`, but ConsoleFormatter formats in a way message looks like this:
![screenshot from 2018-02-19 01-54-21](https://user-images.githubusercontent.com/496233/36359102-d6c20e4c-1517-11e8-9988-8a49efc4cacb.png)
As you see, unnecessary padding looks like anomaly here and there is currently no way to remove it without making own formatter.
This option allows me to replace default padding format with `%s`.
Commits
-------
fbd257c25d [MonologBridge] Allow to change level format
This PR was merged into the 4.1-dev branch.
Discussion
----------
[Lock] Add a TTL to refresh lock
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | see LockInterface's comment
| Tests pass? | yes
| Fixed tickets | NA
| License | MIT
| Doc PR | NA
Using remote locks in long processes needs to defines a fined grain refresh TTL. For instance, when looping over a long list of jobs we can extends the live of the lock by few seconds before processing each item.
But when the the jobs is splitted and each part to take the same time, we can not define the best TTL
Exemple
```
$lock->acquire();
$this->2minutesJob();
$lock->refresh();
$this->5minutesJob();
$lock->refresh();
$this->1minutesJob();
```
The purpose of this PR is to be able to override the default TTL
```
$lock->acquire();
$lock->refresh(120);
$this->2minutesJob();
$lock->refresh(300);
$this->5minutesJob();
$lock->refresh(60);
$this->1minutesJob();
```
Commits
-------
3b1f3286d8 Add a TTL to refresh lock
This PR was squashed before being merged into the 4.1-dev branch (closes#26108).
Discussion
----------
[Serializer] Add a MaxDepth handler
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes <!-- 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 | api-platform/core#1130, api-platform/core#1528, api-platform/core#1528 <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | todo <!--highly recommended for new features-->
Sometimes, instead of just stopping the serialization process when the configured max depth is reached, it can be interesting to let the user return something (like returning the identifier of the entity to stop manually the serialization process).
This PR also makes the max depth handling more similar to circular references handling (that already has the possibility to set a handler).
Commits
-------
ed975c764b [Serializer] Add a MaxDepth handler
This PR was merged into the 4.1-dev branch.
Discussion
----------
Make deprecation notices less verbose
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
I think there is no need to say that deprecated features will be removed in the next major version. That makes messages more verbose for no real reasons.
Commits
-------
0c6ec3fec0 made deprecation notices less verbose
This PR was merged into the 4.1-dev branch.
Discussion
----------
[BrowserKit] add a way to switch to ajax for one request
| Q | A
| ------------- | ---
| Branch? | 4.1
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #20306
| License | MIT
| Doc PR | will do
Follow the work on #20306.
/cc @fabpot is it the right implementation ?
Commits
-------
a10eae7d9e [BrowserKit] add a way to switch to ajax for one request
This PR was merged into the 4.1-dev branch.
Discussion
----------
[PropertyInfo] Added support for extracting type from constructor
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Commits
-------
adcb25ec01 [PropertyInfo] Added support for extracting type from constructor
* 4.0:
[Bridge/Twig] fix composer.json
bug #26086 [FrameworkBundle] Fix using annotation_reader in compiler pass to inject configured cache provider
[WebProfilerBundle] Fix anchor CSS
[HttpKernel] Send new session cookie from AbstractTestSessionListener after session invalidation
[WebProfilerBundle] Tweak default route name
updated StopwatchEvent phpdoc due to the additional of optional float precision introduced in 0db8d7fb6a
Retro-fit proxy code to make it deterministic for older proxy manager implementations
[Serializer] remove unneeded php doc line
Yaml parser regression with comments and non-strings
Fixed broken tests
[TwigBridge] Apply some changes to support Bootstrap4-stable
* 3.4:
[Bridge/Twig] fix composer.json
bug #26086 [FrameworkBundle] Fix using annotation_reader in compiler pass to inject configured cache provider
[WebProfilerBundle] Fix anchor CSS
[HttpKernel] Send new session cookie from AbstractTestSessionListener after session invalidation
[WebProfilerBundle] Tweak default route name
updated StopwatchEvent phpdoc due to the additional of optional float precision introduced in 0db8d7fb6a
Retro-fit proxy code to make it deterministic for older proxy manager implementations
Yaml parser regression with comments and non-strings
Fix undiscoverablility of SymfonyTestsListenerForV7
Fixed broken tests
[TwigBridge] Apply some changes to support Bootstrap4-stable
This PR was merged into the 3.4 branch.
Discussion
----------
[FrameworkBundle] Fix using annotation_reader in compiler pass to inject configured cache provider
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #26086
| License | MIT
| Doc PR | -
The compilation pass of AddAnnotationsCachedReaderPass relies on already removed definition `annotations.cached_reader` due to an alias to `annotation_reader`.
Since the definition is being removed because of alias, configured annotation cache provider is not injected and will default back to ArrayCache.
This PR replaces the use of `annotations.cached_reader` to `annotation_reader` to complete the injection of configured cache provider.
Commits
-------
dfd93da bug #26086 [FrameworkBundle] Fix using annotation_reader in compiler pass to inject configured cache provider
This PR was merged into the 3.4 branch.
Discussion
----------
[HttpKernel] Send new session cookie from AbstractTestSessionListener after session invalidation
When we call `\Symfony\Component\HttpFoundation\Session\Session::invalidate` the session will be emptied and given a new ID, however, since it is empty this `AbstractTestSessionListener` will not send a new cookie to the user, so the user is not caught up to the latest session ID and will re-generate a session with the old session ID on a new visit.
Thus, we the sessionID has changed during a request we must always send a new cookie with the new sessionID, even though the session is empty.
This behaviour is also what is shown in production (non-test) mode.
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | N/A
| License | MIT
Commits
-------
98f5d53 [HttpKernel] Send new session cookie from AbstractTestSessionListener after session invalidation
This PR was merged into the 3.4 branch.
Discussion
----------
[Stopwatch] updated phpdoc due to the addition of optional float precision
| Q | A
| ------------- | ---
| Branch? | 3.4/4.0
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | https://github.com/symfony/symfony/issues/26204
| License | MIT
| Doc PR | -
Commits
-------
8278a47 updated StopwatchEvent phpdoc due to the additional of optional float precision introduced in 0db8d7fb6a
This PR was merged into the 4.1-dev branch.
Discussion
----------
[Process] Allow writing portable "prepared" command lines
| Q | A
| ------------- | ---
| Branch? | 4.1
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #23778
| License | MIT
| Doc PR | symfony/symfony-docs#9295
This give the opportunity to create process commands that allow to changes only the values instead of changing the code.
Commits
-------
d1e4f489d1 [Process] Allow writing portable "prepared" command lines
This PR was merged into the 4.1-dev branch.
Discussion
----------
[Serializer] add a constructor arguement to return csv always as collection
| 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 | #21616
| License | MIT
| Doc PR | TODO create a doc PR for the 3 ways of getting csv collection, or a single
Coding in the train again ;).
![img_9980](https://user-images.githubusercontent.com/3451634/33417042-f13063e4-d59f-11e7-8f30-143da768b1d7.JPG)
This is to be able to add a new behaviour to the csv encoder when passing the alwaysAsCollection context key, this will return a collection even if there is only one element.
Commits
-------
d19d05dc5d [Serializer] add a context key to return csv always as collection
This PR was merged into the 4.1-dev branch.
Discussion
----------
[Serializer] add a context key to return always as collection for XmlEncoder
| Q | A
| ------------- | ---
| Branch? | 4.1
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no <!-- don't forget to update UPGRADE-*.md files -->
| Tests pass? | yes
| Fixed tickets | #25227
| License | MIT
| Doc PR |
This PR add a new `as_collection` context key in order to return always as a collection instead of returning a single elements when there are only one array.
there are only one PR for the CsvEncoder don't wanted to have only One PR containing the two changes. It feel better to have two PR that fix the behaviour on two different things. it's easy to review and to revert if it break something (which should not since we are testing the behaviour).
Commits
-------
adb428d314 [Serializer] add a context key to return always as collection for XmlEncoder
This PR was squashed before being merged into the 4.1-dev branch (closes#26140).
Discussion
----------
[Serializer] deserialize as a null when inner object cannot be created and type hint allows null
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
```php
class ObjectConstructorDummy
{
protected $foo;
public $bar;
private $baz;
public function __construct($foo, $bar, $baz)
{
$this->foo = $foo;
$this->bar = $bar;
$this->baz = $baz;
}
}
class DummyWithNullableConstructorObject
{
private $id;
private $inner;
public function __construct($id, ?ObjectConstructorDummy $inner)
{
$this->id = $id;
$this->inner = $inner;
}
public function getId()
{
return $this->id;
}
public function getInner()
{
return $this->inner;
}
}
```
Trying to deserialize to `DummyWithNullableConstructorObject` with the following data currently fails:
```php
[
'id' => 10,
'inner' => null
]
```
With this PR `DummyWithNullableConstructorObject ` would be constructed with `null` passed as `$inner` because of the type hint.
Commits
-------
2fe9eb1aba [Serializer] deserialize as a null when inner object cannot be created and type hint allows null
This PR was squashed before being merged into the 4.1-dev branch (closes#26213).
Discussion
----------
[FrameworkBundle] Add support to 307/308 HTTP status codes in RedirectController
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #26171
| License | MIT
| Doc PR |
With this PR `RedirectController` will allow to create redirections with use of 307/308 HTTP status codes together with 301/302. Related RFC documents:
* https://tools.ietf.org/html/rfc7231
* https://tools.ietf.org/html/rfc7538
Commits
-------
64fb5a5663 [FrameworkBundle] Add support to 307/308 HTTP status codes in RedirectController
This PR was merged into the 3.4 branch.
Discussion
----------
Retro-fit proxy code to make it deterministic for older proxy manager implementations
Follow up on https://github.com/symfony/symfony/issues/25958#issuecomment-365543535
ProxyManager >= 7.2 already implements a deterministic identifier naming strategy which is critical for reproducible builds (https://github.com/symfony/symfony/issues/25958). but versions below that don’t. This is what this PR fixes. Here is more context: https://github.com/Ocramius/ProxyManager/pull/411
| 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 | -
Commits
-------
0f16056 Retro-fit proxy code to make it deterministic for older proxy manager implementations