* 4.1:
fix cs
fix cs
fix cs
SCA: consolidate non empty array checks across codebase
[cs] correct invalid @param types
[Bridge/PhpUnit] Use composer to download phpunit
[DI] fix taking lazy services into account when dumping the container
[Form] Fixed empty data for compound date interval
[Cache] fix optimizing Psr6Cache for AdapterInterface pools
deal with explicitly enabled workflow nodes
* 3.4:
fix cs
SCA: consolidate non empty array checks across codebase
[cs] correct invalid @param types
[Bridge/PhpUnit] Use composer to download phpunit
[DI] fix taking lazy services into account when dumping the container
[Form] Fixed empty data for compound date interval
[Cache] fix optimizing Psr6Cache for AdapterInterface pools
deal with explicitly enabled workflow nodes
This PR was squashed before being merged into the 3.4 branch (closes#29231).
Discussion
----------
SCA: consolidate non empty array checks across codebase
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
This PR replaces `is_array(...) && count(...) > 0` with `... && is_array(...)` construct used in the codebase.
Commits
-------
2f1fd54dda SCA: consolidate non empty array checks across codebase
This PR was merged into the 3.4 branch.
Discussion
----------
[Cache] fix optimizing Psr6Cache for AdapterInterface pools
| 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 | -.
As described by @Tobion in https://github.com/symfony/symfony/pull/29236/files#r234324045:
> The problem I have experienced is that in dev mode the cache is decorated with a TraceableCache. This means it loses this optimization and introduces https://github.com/symfony/symfony/issues/28918#issuecomment-432399481 again
Commits
-------
b8100a9df0 [Cache] fix optimizing Psr6Cache for AdapterInterface pools
This PR was merged into the 3.4 branch.
Discussion
----------
[DI] fix taking lazy services into account when dumping the container
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #29246
| License | MIT
| Doc PR | -
This PR fixes issues found while working on #29246.
It *does* fix the infinite loop, ~but replaces it by an exception (reopening #29078)~:
> ~It's a requirement to specify a Metadata Driver and pass it to Doctrine\ORM\Configuration::setMetadataDriverImpl()~
The full fix is not immediately accessible as it needs some core changes to the dumping logic. Requiring `symfony/proxy-manager-bridge` works around the issue properly.
See #29251 for 4.2
Commits
-------
67d7623e72 [DI] fix taking lazy services into account when dumping the container
This PR was merged into the 3.4 branch.
Discussion
----------
[Form] Fixed empty data for compound date interval
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #4715 (continuation)
| License | MIT
| Doc PR | ~
Following #29182, since this type has been introduced in Symfony 3 but with the same limitation explained in #4715.
So for consistency this needs to be fixed here as well.
Commits
-------
38a2abc790 [Form] Fixed empty data for compound date interval
This PR was merged into the 4.2-dev branch.
Discussion
----------
[FrameworkBundle] metadata_update_threshold default value must be an int
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no <!-- don't forget to update src/**/CHANGELOG.md files -->
| BC breaks? | no <!-- see https://symfony.com/bc -->
| Deprecations? |no <!-- don't forget to update UPGRADE-*.md and src/**/CHANGELOG.md files -->
| Tests pass? | yes <!-- please add some, will be required by reviewers -->
| Fixed tickets | n/a
| License | MIT
| Doc PR | n/a
The type of the default config value is not in sync with the typehint added to `MetadataBag::__construct()`, so you get an error by default (at least when using strict types):
```
Type error: Argument 2 passed to Symfony\Component\HttpFoundation\Session\Storage\MetadataBag::__construct() must be of the type integer, string given, called in var/cache/test/ContainerEZ62jzd/getSession_Storage_MockFileService.php on line 23 (Behat\Testwork\Call\Exception\FatalThrowableError)
```
Commits
-------
b7e7c46f43 [FrameworkBundle] metadata_update_threshold default value must be an int
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Messenger] Improved message when handler class does not exist
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? |no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
**Problem:**
When defining a non existing messenger handler class in the `services.yml` config file, we encounter this confusing error message:
```
services:
App\Handler\NonExistentHandler:
tags: [messenger.message_handler]
```
```
PHP Fatal error: Uncaught Symfony\Component\Debug\Exception\FatalThrowableError: Argument 1 passed to Symfony\Component\Messenger\DependencyInjection\MessengerPass::guessHandledClasses() must be an instance of ReflectionClass, null given, called in /app/vendor/symfony/messenger/DependencyInjection/MessengerPass.php on line 93 in /app/vendor/symfony/messenger/DependencyInjection/MessengerPass.php:189
Stack trace:
#0 /app/vendor/symfony/messenger/DependencyInjection/MessengerPass.php(93): Symfony\Component\Messenger\DependencyInjection\MessengerPass->guessHandledClasses(NULL, 'App\\Application...')
#1 /app/vendor/symfony/messenger/DependencyInjection/MessengerPass.php(74): Symfony\Component\Messenger\DependencyInjection\MessengerPass->registerHandlers(Object(Symfony\Component\DependencyInjection\ContainerBuilder), Array)
#2 /app/vendor/symfony/dependency-injection/Compiler/Compiler.php(95): Symfony\Component\Messenger\DependencyInjection\MessengerPass->process(Object(Symfony\Component\DependencyInjection\ContainerBuilder))
#3 / in /app/vendor/symfony/messenger/DependencyInjection/MessengerPass.php on line 189
```
**Proposal:**
We can throw a more relevant exception (RuntimeException) in this case to help the developer to have a better understanding of the issue.
```Invalid service "App\Handler\NonExistentHandler": class "App\Handler\NonExistentHandler" does not exist.```
Commits
-------
6ab9274638 Improve error message when defining messenger handler class that does not exists
This PR was merged into the 2.8 branch.
Discussion
----------
[Validator] Added the missing constraints instance checks
| Q | A
| ------------- | ---
| Branch? | 2.8
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
This PR adds the constraints instance checks missing to limit the validators use.
This behavior is already implemented in all built-in validators, but it was missed in two validators.
Commits
-------
0ecaead015 [Validator] Added the missing constraints instance checks
* 4.1:
Undeprecate the single-colon notation for controllers
Command::addOption should allow int in $default
Update symfony links to https
[Form] Fixed keeping hash of equal \DateTimeInterface on submit
[PhpUnitBridge] Fix typo
[Routing] generate(null) should throw an exception
[Form] Minor fixes in docs and cs
[Workflow] Made code simpler
[Config] Unset key during normalization
[Form] Fixed empty data for compound date types
invalidate forms on transformation failures
[FrameworkBundle] fixed guard event names for transitions
method buildTransitionBlockerList returns TransitionBlockerList of expected transition
[FrameworkBundle] fixed guard event names for transitions
[PropertyAccessor] Fix unable to write to singular property using setter while plural adder/remover exist
* 3.4:
Command::addOption should allow int in $default
[Form] Fixed keeping hash of equal \DateTimeInterface on submit
[PhpUnitBridge] Fix typo
[Form] Minor fixes in docs and cs
[Config] Unset key during normalization
[Form] Fixed empty data for compound date types
invalidate forms on transformation failures
[PropertyAccessor] Fix unable to write to singular property using setter while plural adder/remover exist
This PR was merged into the 3.4 branch.
Discussion
----------
[PropertyAccessor] Fix unable to write to singular property using setter while plural adder/remover exist
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #28961
| License | MIT
Please take a look at the tests I added - they describe the issue. It has to do with the priorities: `findAdderAndRemover('User', 'email')` is called earlier than `$this->isMethodAccessible('User', 'setEmail', 1)`
Commits
-------
8238f167ad [PropertyAccessor] Fix unable to write to singular property using setter while plural adder/remover exist
This PR was merged into the 2.8 branch.
Discussion
----------
[Form] Fixed empty data for compound date types
| Q | A
| ------------- | ---
| Branch? | 2.8
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #4715
| License | MIT
| Doc PR | ~
Commits
-------
9bab1e8aba [Form] Fixed empty data for compound date types
This PR was merged into the 4.2-dev branch.
Discussion
----------
[SecurityBundle] Fix remember-me cookie framework inheritance when session is disabled
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #29217
| License | MIT
| Doc PR | N/A
When `framework.session` configuration key is not an array, we ignore it.
Commits
-------
af81008cb6 [SecurityBundle] Fix remember-me cookie framework inheritance when session is disabled
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Translation] make intl+icu format seamless by handling it in MessageCatalogue
| Q | A
| ------------- | ---
| Branch? | 4.2
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #29136
| License | MIT
| Doc PR | -
Commits
-------
c71dfb9673 [Translation] make intl+icu format seamless by handling it in MessageCatalogue
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Messenger] Add handled & sent stamps
| Q | A
| ------------- | ---
| Branch? | 4.2 <!-- see below -->
| Bug fix? | no
| New feature? | yes <!-- don't forget to update src/**/CHANGELOG.md files -->
| BC breaks? | no <!-- see https://symfony.com/bc -->
| Deprecations? | no <!-- don't forget to update UPGRADE-*.md and src/**/CHANGELOG.md files -->
| Tests pass? | yes <!-- please add some, will be required by reviewers -->
| Fixed tickets | N/A <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | symfony/symfony-docs/issues/10661
Based on #29159
This new feature marks sent and handled messages, so middleware can act upon these and use the handler(s) result(s).
This is also the base of a next PR (#29167), introducing a query bus built on top of the message bus.
I'm not sure yet about the best way to determine the handlers and senders names/descriptions to store in the stamps:
- Handlers are callable. I've just reused the [console text descriptor](1c1818b876/src/Symfony/Bundle/FrameworkBundle/Console/Descriptor/TextDescriptor.php (L457-L491)) format for now.
- ~~Sender are `SenderInterface` instances. `\get_class` is used for now, but a single message can be sent by multiple senders, including of the same class.~~ => Updated. Yielding the sender name if provided, the FQCN otherwise.
~~Instead, what about allowing to yield names from locators, and fallback on the above strategies otherwise? So we'll use transport names from the config for senders, and pre-computed compile-time handlers descriptions?~~
=> Done. For handlers, computing it at compile time might not be straightforward. Let's compute it lazily from `HandledStamp::fromCallable()`
---
### From previous conversations:
> What about not adding HandledStamp on `null` returned from handler
IMHO, `null` still is a result. The stamps allows to identify a message as being handled regardless of the returned value, so makes sense on its own and keeping would require one less check for those wanting to consume it.
> What about adding SentStamp?
Makes sense to me and I think it was requested by @Nyholm before on Slack.
So, included in this PR.
> Should it target 4.2 or 4.3?
Targeting 4.2, because of the removal of the handler result forwarding by middleware. A userland middleware could have used this result, typically a cache middleware. Which would now require extra boring code in userland. This will simplify it and allow users to create their query bus instance until 4.3.
Commits
-------
2f5acf790a [Messenger] Add handled & sent stamps
This PR was merged into the 4.1 branch.
Discussion
----------
Undeprecate the single-colon controller notation
| Q | A
| ------------- | ---
| Branch? | 4.1
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a (in core)
| License | MIT
| Doc PR | n/a
This notation is the only way to support controllers as services in the 3.4 LTS version.
This deprecation has only a very small benefit for the Symfony codebase (the amount of code involved is very small), but has a huge cost for the community which cannot avoid this deprecation without dropping support for the LTS or making crazy logic to switch routing files (as they cannot switch things inline in YAML or XML files).
This deprecation will be delayed until a future 5.x version, when the current LTS will be 4.4 (which supports the new notation).
This PR is based on a discussion I had with @nicolas-grekas a few days ago about the cost of that deprecation for the community.
For instance, in FOSUserBundle, a [PR](https://github.com/FriendsOfSymfony/FOSUserBundle/pull/2853) got merged which duplicates the routing files with new files using the new notation, allowing people to opt-out from the deprecation, while the original files cannot be migrated without dropping support for the LTS (which is bad for the concept of LTS, as its usefulness is severally impacted if the ecosystem does not support it). But that's a bad solution for the future, as the bundle would have to keep 2 files forever (well, until next major) to avoid breaking projects referencing the new files. And most open-source bundles providing routes are faced with this challenge.
This is sent to 4.1, so that projects using 4.1 can become deprecation-free for these community bundles when using the latest 4.1 release too (meaning FOSUserBundle could revert this PR before the next release for instance)
Commits
-------
453efdfe1e Undeprecate the single-colon notation for controllers
