This PR was merged into the 3.4 branch.
Discussion
----------
[HttpKernel] Fix dealing with self/parent in ArgumentMetadataFactory
| 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 | -
Applies CS fixes that where merged on 4.0 to 3.4, embeds a fix in ArgumentMetadataFactory, which couldn't deal with self/parent type hints.
Commits
-------
ba5cb1a245 fixed CS
* 3.4:
fix merge
[Security] Fix logout
Cleanup 2 tests for the HttpException classes
#27250 limiting GET_LOCK key up to 64 char due to changes in MySQL 5.7.5 and later
[Config] Fix tests when path contains UTF chars
[DI] Shared services should not be inlined in non-shared ones
[Profiler] Remove propel & event_listener_loading category identifiers
[Filesystem] Fix usages of error_get_last()
[Cache][Lock] Fix usages of error_get_last()
[Debug] Fix populating error_get_last() for handled silent errors
[DI] Display previous error messages when throwing unused bindings
Suppress warnings when open_basedir is non-empty
* 2.8:
[Security] Fix logout
#27250 limiting GET_LOCK key up to 64 char due to changes in MySQL 5.7.5 and later
[Profiler] Remove propel & event_listener_loading category identifiers
[Filesystem] Fix usages of error_get_last()
[Debug] Fix populating error_get_last() for handled silent errors
Suppress warnings when open_basedir is non-empty
* 2.7:
[Security] Fix logout
#27250 limiting GET_LOCK key up to 64 char due to changes in MySQL 5.7.5 and later
[Profiler] Remove propel & event_listener_loading category identifiers
[Filesystem] Fix usages of error_get_last()
[Debug] Fix populating error_get_last() for handled silent errors
Suppress warnings when open_basedir is non-empty
This PR was squashed before being merged into the 2.7 branch (closes#24805).
Discussion
----------
[Security] Fix logout
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | no
| Fixed tickets | #6751, #7104
| License | MIT
Commits
-------
9e88eb5aa9 [Security] Fix logout
This PR was merged into the 2.7 branch.
Discussion
----------
[Process] Suppress warnings when open_basedir is non-empty
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
If PHP is configured *with a non-empty open_basedir* value that does not permit access to the target location, these calls to is_executable() throw warnings.
While Symfony may not raise exceptions for warnings in production environments, other frameworks (such as Laravel) do, in which case any of these checks causes a show-stopping 500 error.
We fixed a similar issue in the ExecutableFinder class via symfony/symfony#16182 .
This has always been an issue, but 709e15e7a3 made it more likely that a warning is triggered.
Commits
-------
34f136e01b Suppress warnings when open_basedir is non-empty
This PR was submitted for the master branch but it was merged into the 3.4 branch instead (closes#27256).
Discussion
----------
Cleanup 2 tests for the HttpException classes
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #27238
| License | MIT
| Doc PR | -
See discussion in linked issue.
Commits
-------
5516b329af Cleanup 2 tests for the HttpException classes
This PR was squashed before being merged into the 4.1 branch (closes#27128).
Discussion
----------
[Messenger] Middleware factories support in config
| Q | A
| ------------- | ---
| Branch? | master <!-- 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 | todo
Following https://github.com/symfony/symfony/pull/26864, this would allow to configure easily the middlewares by using an abstract factory definition to which are provided simple arguments (just scalars, no services references).
For instance, here is how the DoctrineBundle would benefit from such a feature (also solving the wiring of the `DoctrineTransactionMiddleware` reverted in https://github.com/symfony/symfony/pull/26684):
```yaml
framework:
messenger:
buses:
default:
middleware:
- logger
- doctrine_transaction_middleware: ['entity_manager_name']
```
where `doctrine_transaction_middleware` would be an abstract factory definition provided by the doctrine bundle:
```yml
services:
doctrine.orm.messenger.middleware_factory.transaction:
class: Symfony\Bridge\Doctrine\Messenger\DoctrineTransactionMiddlewareFactory
arguments: ['@doctrine']
doctrine_transaction_middleware:
class: Symfony\Bridge\Doctrine\Messenger\DoctrineTransactionMiddleware
factory: ['@doctrine.orm.messenger.middleware_factory.transaction', 'createMiddleware']
abstract: true
# the default arguments to use when none provided from config.
# i.e:
# middlewares:
# - doctrine_transaction_middleware: ~
arguments: ['default']
```
and is interpreted as:
```yml
buses:
default:
middleware:
-
id: logger
arguments: { }
-
id: doctrine_transaction_middleware
arguments:
- entity_manager_name
default_middleware: true
```
---
<details>
<summary>Here is the whole config reference with these changes: </summary>
```yaml
# Messenger configuration
messenger:
enabled: true
routing:
# Prototype
message_class:
senders: []
serializer:
enabled: true
format: json
context:
# Prototype
name: ~
encoder: messenger.transport.serializer
decoder: messenger.transport.serializer
adapters:
# Prototype
name:
dsn: ~
options: []
default_bus: null
buses:
# Prototype
name:
default_middleware: true
middleware:
# Prototype
-
id: ~ # Required
arguments: []
```
</details>
Commits
-------
f5ef421474 [Messenger] Middleware factories support in config
This PR was merged into the 4.1 branch.
Discussion
----------
[HttpKernel] Fix services are no longer injected into __invoke controllers method
| Q | A
| ------------- | ---
| Branch? | 4.1 <!-- see below -->
| 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 | #27208 <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | N/A
_TL;DR:_ The `RemoveEmptyControllerArgumentLocatorsPass` is the one adding the `Controller::_invoke` => `Controller` shortcut missing from the service locator. It isn't properly executed on some cases. This fixes it.
Since https://github.com/symfony/symfony/pull/26833, the resolvers are decorated by a `TraceableValueResolver`, which usually isn't much an issue to deal within passes. But the `RemoveEmptyControllerArgumentLocatorsPass` happens late (`TYPE_BEFORE_REMOVING`), when decoration inheritance is already resolved, so accessing `$controllerLocator = $container->getDefinition((string) $serviceResolver->getArgument(0));` isn't accessing the controller locator, but the decorated service instead.
Commits
-------
ee44903fd0 [HttpKernel] Fix services are no longer injected into __invoke controllers method
This PR was merged into the 2.7 branch.
Discussion
----------
[Debug] Fix populating error_get_last() for handled silent errors
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
When a userland error handler doesn't return `false`, `error_get_last()` is not updated, so we cannot see the real last error, but the previous one.
See https://3v4l.org/Smmt7
Commits
-------
d7e612d2ac [Debug] Fix populating error_get_last() for handled silent errors
This PR was merged into the 4.1 branch.
Discussion
----------
[HttpKernel] Make TraceableValueResolver $stopwatch mandatory
| Q | A
| ------------- | ---
| Branch? | 4.1 <!-- see below -->
| Bug fix? | no
| 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 | https://github.com/symfony/symfony/pull/26833#discussion_r181997515 <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | N/A
I understand why this was suggested in https://github.com/symfony/symfony/pull/26833#discussion_r181997515, but as stated by @iltar , I don't think it makes sense to register a traceable resolver instantiating a Stopwatch itself, as there is no way to fetch it and wouldn't be a shared instance, probably defeating the feature and registering a useless decorator.
Instead, let's make the stopwatch mandatory and make the service id to use in the pass configurable.
Commits
-------
585ae7c646 [HttpKernel] Make TraceableValueResolver $stopwatch mandatory
This PR was merged into the 3.4 branch.
Discussion
----------
[Cache][Lock] Fix usages of error_get_last()
| 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 | -
When a userland error handler doesn't return `false`, `error_get_last()` is not updated, so we cannot see the real last error, but the previous one.
See https://3v4l.org/Smmt7
Commits
-------
7904784a94 [Cache][Lock] Fix usages of error_get_last()
This PR was merged into the 2.7 branch.
Discussion
----------
[Filesystem] Fix usages of error_get_last()
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Same as #27232 for 2.7.
When a userland error handler doesn't return `false`, `error_get_last()` is not updated, so we cannot see the real last error, but the previous one.
See https://3v4l.org/Smmt7
Commits
-------
9d015c7c50 [Filesystem] Fix usages of error_get_last()
This PR was squashed before being merged into the 4.1 branch (closes#27190).
Discussion
----------
[Workflow] Added DefinitionBuilder::setMetadataStore().
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
This PR complements #26092.
Commits
-------
2882f8d8c8 [Workflow] Added DefinitionBuilder::setMetadataStore().