This PR was merged into the 3.4 branch.
Discussion
----------
[DI] Add "container.hot_path" tag to flag the hot path and inline related services
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
This PR is the result of my quest to squeeze some performance out of 3.4/4.0.
It builds on two ideas:
- a new `container.inline` tag that identifies the services that are *always* needed. This tag is only applied to a very short list of bootstrapping services (`router`, `event_dispatcher`, `http_kernel` and `request_stack` only). Then, it is propagated to all dependencies of these services, with a special case for event listeners, where only listed events are propagated to their related listeners.
- replacing the PHP autoloader by plain inlined `require_once` in generated service factories, with the benefit of completely bypassing the autoloader for services and their class hierarchy.
The end result is significant, even on a simple Hello World.
Here is the Blackfire profile, results are consistent with `ab` benchmarks:
https://blackfire.io/profiles/compare/b5fa5ef0-755c-4967-b990-572305f8f381/graph
![capture du 2017-11-08 16-54-28](https://user-images.githubusercontent.com/243674/32558666-a3f439b2-c4a5-11e7-83a3-db588c3e21e5.png)
Commits
-------
f7cb559a06 [DI] Add "container.hot_path" tag to flag the hot path and inline related services
* 3.3:
Random fixes
Docblock fixes
[HttpKernel] Enhance deprecation message
[SecurityBundle] Fix the datacollector to properly support decision.object being null
This PR was merged into the 3.3 branch.
Discussion
----------
[SecurityBundle] Fix the datacollector to properly support decision.object being null
| Q | A
| ------------- | ---
| Branch? | 3.3 <!-- see comment below -->
| 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 | #24804 <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | N/A
Similar to https://github.com/symfony/symfony/pull/23050, when accessing a `Data` clone property through `__get()` and the value is `null` for instance, you'll really get `null` instead of a `Data` instance. The solution is to use `seek` instead whenever we access and try to use `profiler_dump` on a `Data` property that can be a simple scalar like `null` or `false`. AFAIK, `decision.object` is the only one here.
Commits
-------
769a5f204f [SecurityBundle] Fix the datacollector to properly support decision.object being null
This PR was merged into the 2.8 branch.
Discussion
----------
Fix dump panel hidden when closing a dump
| Q | A
| ------------- | ---
| Branch? | 2.8
| Bug fix? | yes-ish
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
In the dump panel of the debug bar, when closing a dump the panel sometimes get hidden:
![before](https://user-images.githubusercontent.com/1736542/31867025-615e9c48-b788-11e7-8329-96716c211523.gif)
This is because when the size of the panel is reduced, if the mouse is not over it anymore, the `:hover` pseudo-class does not apply anymore.
I "fixed" it by setting a min-height on the panel when closing a dump. The min-height is removed when leaving the panel _on purpose_:
![after](https://user-images.githubusercontent.com/1736542/31867054-d01a01cc-b788-11e7-9ef7-8418ae2b3094.gif)
For now I only tested it on Firefox 56 on Arch Linux.
Commits
-------
2e0b263d9c Fix dump panel hidden when closing a dump
* 3.3:
[Serializer] Fix extra attributes when no group specified
[Intl] Make intl-data tests pass and save language aliases again
[Console] Fix CommandTester::setInputs() docblock
[Serializer] readd default argument value
[VarDumper] fix trailling comma when dumping an exception
Remove useless docblocks
[FrameworkBundle] Fix docblocks
[PropertyInfo] Remove useless docblocks
This PR was squashed before being merged into the 2.7 branch (closes#24666).
Discussion
----------
[WebProfilerBundle] add missing tests
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | none
| License | MIT
This PR fixes one test in WebDebugToolbarListenerTest and adds one that was missing (more detailed description is available in the commits description)
Commits
-------
363d3a8cf2 [WebProfilerBundle] add missing tests
This PR was merged into the 2.7 branch.
Discussion
----------
[FrameworkBundle][Routing] Remove unused logger argument
| Q | A
| ------------- | ---
| Branch? | 2.7
| Bug fix? | no
| 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 | N/A <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | N/A
This argument was never used as far as the FrameworkBundle Router was.
But actually don't we want to add the `$logger` argument to this class? It's used by the `UrlGenerator` when `router.strict_requirements` is `false` for instance:
<img width="1064" alt="screenshot 2017-10-29 a 09 57 31" src="https://user-images.githubusercontent.com/2211145/32142080-482bc64e-bc90-11e7-8382-b78b507bae48.PNG">
Commits
-------
dc9492eb6f [FrameworkBundle][Routing] Remove unused logger argument
This PR was merged into the 3.4 branch.
Discussion
----------
symfony/form auto-enables symfony/validator, even when not present
| Q | A
| ------------- | ---
| Branch? | 3.4 or master / 2.7, 2.8 or 3.3 <!-- see comment below -->
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | n/a
| License | MIT
| Doc PR | not needed
In #24303, we allowed form to be used without the validator component. But, there is a small problem with the logic: the validation system is set to enabled, even if it is not present. If you install form but NOT validator, you see the error:
> Validation support cannot be enabled as the Validator component is not installed.
Assuming the form system really is usable without validation, this should be an easy merge.
Commits
-------
03c0254296 Only enabling validation if it is present
This PR was merged into the 3.4 branch.
Discussion
----------
[HttpKernel] Move services reset to Kernel::handle()+boot()
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #24552
| License | MIT
| Doc PR | -
This is an alternative to #24697 (which uses middlewares).
This PR adds a new `services_resetter` service that the Kernel calls on 2nd root requests to reset services.
Instead of #24697 which plans for optional enabling of the services reset, this approach moves the responsibility of calling the services resetter to the core Kernel class, so that no configuration/middleware/etc. is required at all, and no overhead exists at all for regular requests.
Commits
-------
4501a3688b [HttpKernel] Move services reset to Kernel
* 3.3:
Fixing a bug where non-existent classes would cause issues
[SecurityBundle] hotfix: update phpdocs on logout url
[FrameworkBundle] Do not load property_access.xml if the component isn't installed
Fixed a few spelling mistakes in Luxembourgish translation
* 2.8:
[SecurityBundle] hotfix: update phpdocs on logout url
[FrameworkBundle] Do not load property_access.xml if the component isn't installed
Fixed a few spelling mistakes in Luxembourgish translation
This PR was squashed before being merged into the 2.7 branch (closes#24605).
Discussion
----------
[FrameworkBundle] Do not load property_access.xml if the component isn't installed
| Q | A
| ------------- | ---
| Branch? | 2.7 <!-- see comment below -->
| 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 | https://github.com/symfony/symfony/pull/24563#issuecomment-337549147 <!-- #-prefixed issue number(s), if any -->
| License | MIT
| Doc PR | N/A
This PR actually aims to fix https://github.com/symfony/symfony/pull/24563#issuecomment-337549147 as the exception introduced in the PR can't be reached anyway when using the FrameworkBundle without the property access component as you'll get:
> Uncaught Symfony\Component\Debug\Exception\ClassNotFoundException: Attempted to load class "PropertyAccessor" from namespace "Symfony\Component\PropertyAccess".
With this fix, you properly get:
> The ObjectNormalizer class requires the "PropertyAccess" component. Install "symfony/property-access" to use it.
Not sure this change really belongs to a patch release, but the original PR was accepted in the 2.7 branch.
Also, I'd rather remove the ObjectNormalizer definition if the component isn't available, as suggested by @xabbuh (https://github.com/symfony/symfony/pull/24563#issuecomment-336795644). But in 2.7, this is the only normalizer registered by default and the [`SerializerPass` throws an exception if no normalizer is registered.](https://github.com/symfony/symfony/blob/2.7/src/Symfony/Bundle/FrameworkBundle/DependencyInjection/Compiler/SerializerPass.php#L46)
To sum up, either:
1. we completly prevent using the FrameworkBundle and the serializer without the property access component, even if you don't really care about the ObjectNormalizer because you only use your owns specific ones. (and you'll get the exception hinting to install the property access component)
2. we allow using the FrameworkBundle and the serializer without the property access component, so we remove the ObjectNormalizer definition, but the user'll get a `You must tag at least one service as "serializer.normalizer" to use the Serializer service` exception until he configures a normalizer (and we don't get the hint about installing the property access component to enable the ObjectNormalizer. We already have a suggest entry in the composer.json file, though).
To me option 2 looks better. WDYT?
Commits
-------
d297e27600 [FrameworkBundle] Do not load property_access.xml if the component isn't installed
* 3.3:
Ensure DeprecationErrorHandler::collectDeprecations() is triggered
[FrameworkBundle] Allow to disable assets via framework:assets xml configuration
fixed $_ENV/$_SERVER precedence in test framework
[HttpFoundation] Fix FileBag issue with associative arrays
[DI] Throw when a service name or an alias contains dynamic values (prevent an infinite loop)
fix the phpdoc that is not really inherited from response
Minor docblock cleanup
Remove redundant sprintf arguments.
* 2.8:
[HttpFoundation] Fix FileBag issue with associative arrays
fix the phpdoc that is not really inherited from response
Minor docblock cleanup
Remove redundant sprintf arguments.
* 2.7:
[HttpFoundation] Fix FileBag issue with associative arrays
fix the phpdoc that is not really inherited from response
Minor docblock cleanup
Remove redundant sprintf arguments.