* 4.1:
[VarDumper] fix dump of closures created from callables
[DI] fix dumping inlined services
Add framework asset changes to upgrade 3.0 guide
[Travis] Bump ext-mongodb to 1.5.2 on Travis
[DI] dont track classes/interfaces used to compute autowiring error messages
[DI] fix GraphvizDumper ignoring inline definitions
bumped Symfony version to 4.1.8
updated VERSION for 4.1.7
updated CHANGELOG for 4.1.7
bumped Symfony version to 3.4.19
updated VERSION for 3.4.18
updated CHANGELOG for 3.4.18
bumped Symfony version to 2.8.48
updated VERSION for 2.8.47
update CONTRIBUTORS for 2.8.47
updated CHANGELOG for 2.8.47
Fix ini_get() for boolean values
* 3.4:
[VarDumper] fix dump of closures created from callables
[DI] fix dumping inlined services
Add framework asset changes to upgrade 3.0 guide
[Travis] Bump ext-mongodb to 1.5.2 on Travis
[DI] dont track classes/interfaces used to compute autowiring error messages
[DI] fix GraphvizDumper ignoring inline definitions
bumped Symfony version to 3.4.19
updated VERSION for 3.4.18
updated CHANGELOG for 3.4.18
bumped Symfony version to 2.8.48
updated VERSION for 2.8.47
update CONTRIBUTORS for 2.8.47
updated CHANGELOG for 2.8.47
Fix ini_get() for boolean values
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] added support for connecting to Redis clusters via DSN
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Replaces #28300 and #28175
This PR allows configuring a cluster of Redis servers using all available options of either the phpredis extension or the Predis package:
- the `redis_cluster=0/1` boolean option configures whether the client should use the Redis cluster protocol;
- several hosts can be provided using a syntax very similar to #28598, enabling consistent hashing distribution of keys;
- `failover=error/distribute/slaves` can be set to direct reads at slave servers;
- extra options are passed as is to the driver (e.g. `profile=2.8`)
- Predis per-server settings are also possible, using e.g. `host[localhost][alias]=foo` in the query string, or `host[localhost]=alias%3Dfoo` (ie PHP query arrays or urlencoded key/value pairs)
Commits
-------
a42e8774d6 [Cache] added support for connecting to Redis clusters via DSN
This PR was merged into the 4.2-dev branch.
Discussion
----------
[FrameworkBundle] Moving Cache-related CompilerPass to Cache component
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | yes
| Deprecations? | yes
| Tests pass? | yes
| Fixed tickets |
| License | MIT
| Doc PR |
Hi, first PR here 🎉
This is related to #27479 and a first work to move Cache-related CompilerPass out of `FrameworkBundle`, it allows to decouple part of the FrameworkBundle configuration classes.
Since we didn't choosed a fixed directory organization to manage theses, I actually did same as in Bundles and used DependencyInjection folder. If we do choose to follow my [last comment directory organization proposal](https://github.com/symfony/symfony/issues/27479#issuecomment-401307685), I'll move theses CompilerPass to `Framework/DependencyInjection/Compiler` directory (nothing hard here).
Thanks to @DanieleGBX that helped me checking this PR and gave me some good advices !
Here is a list of all CompilerPass I moved, with related component (I also moved related tests when they were present):
- **Cache** - CacheCollectorPass
- **Cache** - CachePoolClearerPass
- **Cache** - CachePoolPass
- **Cache** - CachePoolPrunerPass
Commits
-------
53e7040829 moving Cache-related compiler pass from FrameworkBundle to Cache component
* 4.1:
[Console] simplified code
removed useless phpdoc
improve docblocks around group sequences
[Cache] prevent getting older entries when the version key is evicted
[WebProfilerBundle] added a note in the README
[Yaml] Skip parser test with root user
[Filesystem] Skip tests on readable file when run with root user
[FWBundle] Fix an error in WebTestCase::createClient's PHPDoc
[HttpFoundation][Security] forward locale and format to subrequests
[Console] Send the right exit code to console.terminate listeners
[HttpFoundation] fix hidding warnings from session handlers
Caching missed templates on cache warmup
* 3.4:
[Console] simplified code
removed useless phpdoc
improve docblocks around group sequences
[Cache] prevent getting older entries when the version key is evicted
[WebProfilerBundle] added a note in the README
[Yaml] Skip parser test with root user
[Filesystem] Skip tests on readable file when run with root user
[FWBundle] Fix an error in WebTestCase::createClient's PHPDoc
[HttpFoundation][Security] forward locale and format to subrequests
[Console] Send the right exit code to console.terminate listeners
[HttpFoundation] fix hidding warnings from session handlers
Caching missed templates on cache warmup
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] add "setCallbackWrapper()" on adapters implementing CacheInterface for more flexibility
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #27730
| License | MIT
| Doc PR | -
Preparing my talk at SymfonyLive London, see you all there :)
This allows wrapping the callback passed to `->get($item, $callback, $beta)` in a callable that should at least return `$callback($item)`, but can do something around the call.
The default wrapper is locking the key to provide lock-based stampede protection.
That was already the case before this PR, but in a much dirtier way at the design level.
Fixes a few issues found meanwhile.
Commits
-------
8cf3625b11 [Cache] add "setCallbackWrapper()" on adapters implementing CacheInterface for more flexibility
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] improve perf when using RedisCluster by reducing roundtrips to the servers
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
Improves perf when using RedisCluster by:
- disabling versioning and replacing by per-master clear/flush
- grouping multiple "get" in one "mget"
- enabling pipelining in PredisCluster mode
~Might need adjustment depending on the answer to https://github.com/nrk/predis/issues/520~
Commits
-------
5155f48029 [Cache] improve perf when using RedisCluster by reducing roundtrips to the servers
* 4.1:
bumped Symfony version to 2.8.46
updated VERSION for 2.8.45
update CONTRIBUTORS for 2.8.45
updated CHANGELOG for 2.8.45
[PhpUnitBridge] keep compat with composer 1.0
Instantiate $offset and $maxlen at definition
[Cache] minor code update to leverage PHP 7.1
This PR was merged into the 4.2-dev branch.
Discussion
----------
[VarExporter] a new component to serialize values to plain PHP code
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
This PR proposes moving what is currently the `PhpMarshaller` class in the Cache component to a separate component.
This component would provide only one public static method:
`VarExporter::export($value, bool &$isStaticValue = null): string`.
This method returns `$value` serialized as plain PHP code. Running this code creates the same exact data structure that `$value` contained. This is exactly like `serialize()` and `unserialize()`, from which all semantics are preserved (`__sleep`, `__wakeup` and `Serializable`).
The reason to use this method *vs* `serialize()` or even igbinary is performance: thanks to OPcache, the resulting code is significantly faster and more memory efficient than using `unserialize()` or `igbinary_unserialize()`.
Unlike `var_export()`, this works on any serializable PHP value.
It also provides a few improvements over `var_export()`/`serialize()`:
- the output is PSR-2 compatible
- the output can be re-indented without messing up with any `\r` or `\n` in the data
- missing classes throw a `ReflectionException` instead of being unserialized to a `PHP_Incomplete_Class` object
- references involving `SplObjectStorage`, `ArrayObject` or `ArrayIterator` instances are preserved
- `Reflection*`, `IteratorIterator` and `RecursiveIteratorIterator` classes throw an exception when being serialized (their unserialized version is broken anyway, see https://bugs.php.net/76737.)
Commits
-------
7831ad75e5 [VarExporter] a new component to serialize values to plain PHP code
* 4.1:
[DoctrineBridge] support __toString as documented for UniqueEntityValidator
[travis] enable Redis cluster
[Cache] enable Memcached::OPT_TCP_NODELAY to fix perf of misses
fix data mapper return type in docblock
fix type error handling when writing values
* 3.4:
[DoctrineBridge] support __toString as documented for UniqueEntityValidator
[travis] enable Redis cluster
[Cache] enable Memcached::OPT_TCP_NODELAY to fix perf of misses
fix data mapper return type in docblock
fix type error handling when writing values
This PR was merged into the 3.4 branch.
Discussion
----------
[Cache] enable Memcached::OPT_TCP_NODELAY to fix perf of misses
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #28248
| License | MIT
| Doc PR | -
As described in https://github.com/php-memcached-dev/php-memcached/issues/24 and because we enable the binary protocol by default.
Commits
-------
8b59d177db [Cache] enable Memcached::OPT_TCP_NODELAY to fix perf of misses
* 4.1:
Fix Clidumper tests
Enable the fixer enforcing fully-qualified calls for compiler-optimized functions
Apply fixers
Disable the native_constant_invocation fixer until it can be scoped
Update the list of excluded files for the CS fixer
* 4.0:
Fix Clidumper tests
Enable the fixer enforcing fully-qualified calls for compiler-optimized functions
Apply fixers
Disable the native_constant_invocation fixer until it can be scoped
Update the list of excluded files for the CS fixer
* 3.4:
Fix Clidumper tests
Enable the fixer enforcing fully-qualified calls for compiler-optimized functions
Apply fixers
Disable the native_constant_invocation fixer until it can be scoped
Update the list of excluded files for the CS fixer
* 2.8:
Fix Clidumper tests
Enable the fixer enforcing fully-qualified calls for compiler-optimized functions
Apply fixers
Disable the native_constant_invocation fixer until it can be scoped
Update the list of excluded files for the CS fixer
This PR was merged into the 4.2-dev branch.
Discussion
----------
Add symfony/contracts: a set of abstractions extracted out of the Symfony components
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | -
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
A set of abstractions extracted out of the Symfony components.
This is a topic I've been thinking about for a long time. I feel like the time has come for Symfony to publish some abstractions so that people could build on them in a decoupled way.
I've identified interfaces in some components that would greatly benefit from being moved out from the components where they are for now. E.g. #26929 is something that has a broader scope than the Cache component itself.
By putting them in a new `symfony/abstractions` package, we would allow more innovation in the Symfony community, at the abstraction level.
In order to start small, I propose only one interface that gathers a concept that is shared amongst many components already: `ResetInterface`. It would provide a standard `reset()` method, whose purpose is to set an object back to its initial state, allowing it to be reused many times with no side effects/leaks related to its history. By this definition, it could also be autoconfigured (as done here, see update in FrameworkExtension). See wording in the docblock in the attached source code.
Ideally, I'd like this package to provide not only interfaces, by also generic traits, and reference test suites when possible. We could work on adding more abstractions during the 4.2 cycle. WDYT?
## Here is the attached README:
A set of abstractions extracted out of the Symfony components.
Can be used to build on semantics that the Symfony components proved useful - and
that already have battle tested implementations.
Design Principles
-----------------
* contracts are split by domain, each into their own sub-namespaces;
* contracts are small and consistent sets of PHP interfaces, traits, normative
docblocks and reference test suites when applicable, etc.;
* all contracts must have a proven implementation to enter this repository;
* they must be backward compatible with existing Symfony components.
FAQ
---
### How to use this package?
The abstractions in this package are useful to achieve loose coupling and
interoperability. By using the provided interfaces as type hints, you are able
to reuse any implementations that match their contracts. It could be a Symfony
component, or another one provided by the PHP community at large.
Depending on their semantics, some interfaces can be combined with autowiring to
seamlessly inject a service in your classes.
Others might be useful as labeling interfaces, to hint about a specific behavior
that could be enabled when using autoconfiguration or manual service tagging (or
any other means provided by your framework.)
### How is this different from PHP-FIG's PSRs?
When applicable, the provided contracts are built on top of PHP-FIG's PSR. We
encourage relying on them and won't duplicate the effort. Still, the FIG has
different goals and different processes. Here, we don't need to seek universal
standards. Instead, we're providing abstractions that are compatible with the
implementations provided by Symfony. This should actually also contribute
positively to the PHP-FIG (from which Symfony is a member), by hinting the group
at some abstractions the PHP world might like to take inspiration from.
### Why isn't this package split into several packages?
Putting all interfaces in one package eases discoverability and dependency
management. Instead of dealing with a myriad of small packages and the
corresponding matrix of versions, you just need to deal with one package and one
version. Also when using IDE autocompletion or just reading the source code, it
makes it easier to figure out which contracts are provided.
There are two downsides to this approach: you may have unused files in your
`vendor/` directory, and in the future, it will be impossible to use two
different sub-namespaces in different major versions of the package. For the
"unused files" downside, it has no practical consequences: their file sizes are
very small, and there is no performance overhead at all since they are never
loaded. For major versions, this package follows the Symfony BC + deprecation
policies, with an additional restriction to never remove deprecated interfaces.
Resources
---------
* [Documentation](https://symfony.com/doc/current/components/contracts.html)
* [Contributing](https://symfony.com/doc/current/contributing/index.html)
* [Report issues](https://github.com/symfony/symfony/issues) and
[send Pull Requests](https://github.com/symfony/symfony/pulls)
in the [main Symfony repository](https://github.com/symfony/symfony)
Commits
-------
898203649f Added symfony/contracts: a set of abstractions extracted out of the components
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] Add `MarshallerInterface` allowing to change the serializer, providing a default one that automatically uses igbinary when available
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | yes
| Tests pass? | yes
| Fixed tickets | #19895
| License | MIT
| Doc PR | -
With this PR, when igbinary is available, it is automatically used to serialize values.
This provides faster and smaller cache payloads.
The unserializing logic is autoadaptative:
- when an igbinary-serialized value is unserialized but the extension is missing, a cache miss is triggered
- when a natively-serialized value is unserialized and the extension is available, the native `unserialize()` is used
Ping @palex-fpt since you provided very useful comments on the topic and might be interested in reviewing here also.
Commits
-------
9c328c4894 [Cache] Add `MarshallerInterface` allowing to change the serializer, providing a default one that automatically uses igbinary when available
This PR was merged into the 4.2-dev branch.
Discussion
----------
[FrameworkBundle][Cache] Allow configuring PDO-based cache pools, with table auto-creation on first use
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
* Allowed configuring PDO-based cache pools via a new `cache.adapter.pdo` abstract service
* added automatic table creation when using Doctrine DBAL with PDO-based backends
Commits
-------
1484117430 [FrameworkBundle][Cache] Allow configuring PDO-based cache pools, with table auto-creation on first use
* 4.1:
Add color support for Hyper terminal .
[HttpFoundation] Fix tests: new message for status 425
[Doctrine Bridge] Fixed usage of wrong variable when tagged subscriber is invalid
[Workflow] Update phpdoc to fit a used className
[PropertyInfo] added handling of nullable types in PhpDoc
[HttpKernel] Make AbstractTestSessionListener compatible with CookieClearingLogoutHandler
[Cache] provider does not respect option maxIdLength with versioning enabled
[Form] Fix fixtures for forward compat
[Lock] Fix SemaphoreStoreTest on OS X
Ensure the class discriminator mechanism works with serialization groups as well
fix handling of empty DI extension configs
* 4.0:
Add color support for Hyper terminal .
[HttpFoundation] Fix tests: new message for status 425
[Doctrine Bridge] Fixed usage of wrong variable when tagged subscriber is invalid
[PropertyInfo] added handling of nullable types in PhpDoc
[HttpKernel] Make AbstractTestSessionListener compatible with CookieClearingLogoutHandler
[Cache] provider does not respect option maxIdLength with versioning enabled
* 3.4:
Add color support for Hyper terminal .
[HttpFoundation] Fix tests: new message for status 425
[Doctrine Bridge] Fixed usage of wrong variable when tagged subscriber is invalid
[PropertyInfo] added handling of nullable types in PhpDoc
[HttpKernel] Make AbstractTestSessionListener compatible with CookieClearingLogoutHandler
[Cache] provider does not respect option maxIdLength with versioning enabled
* 4.1:
[minor] SCA
[Serializer] Minor tweaks for a67b650f12
allow_extra_attributes does not throw an exception as documented
[Cache] fix visibility of RedisTrait::init()
[Serializer] Updates DocBlock to a mixed param type
* 4.0:
[minor] SCA
[Serializer] Minor tweaks for a67b650f12
allow_extra_attributes does not throw an exception as documented
[Cache] fix visibility of RedisTrait::init()
[Serializer] Updates DocBlock to a mixed param type
* 3.4:
[Serializer] Minor tweaks for a67b650f12
allow_extra_attributes does not throw an exception as documented
[Cache] fix visibility of RedisTrait::init()
[Serializer] Updates DocBlock to a mixed param type
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] Fix locking on Solaris
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
As reported on #27668, the descriptor must be writeable on Solaris to get an exclusive lock.
Commits
-------
43da583267 [Cache] Fix locking on Solaris
* 4.1:
[Lock] use 'r+' for fopen (fixes issue on Solaris)
[HttpKernel] fix test compat with PHP 5.3
fix handling of nested Error instances
fix file lock on SunOS
Ignore keepQueryParams attribute when generating route redirect.
[Cache] more granular handling of exceptions in AbstractTrait::clear()
change `evaluate()` docblock return type from string to mixed
Set serialize_precision explicitly to avoid fancy float rounding
* 4.0:
[Lock] use 'r+' for fopen (fixes issue on Solaris)
[HttpKernel] fix test compat with PHP 5.3
fix handling of nested Error instances
fix file lock on SunOS
[Cache] more granular handling of exceptions in AbstractTrait::clear()
change `evaluate()` docblock return type from string to mixed
Set serialize_precision explicitly to avoid fancy float rounding
* 3.4:
[Lock] use 'r+' for fopen (fixes issue on Solaris)
[HttpKernel] fix test compat with PHP 5.3
fix handling of nested Error instances
fix file lock on SunOS
[Cache] more granular handling of exceptions in AbstractTrait::clear()
change `evaluate()` docblock return type from string to mixed
Set serialize_precision explicitly to avoid fancy float rounding
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] added support for phpredis 4 `compression` and `tcp_keepalive` options
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | yes
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
See https://pecl.php.net/package-changelog.php?package=redis
Commits
-------
2ff02cd333 [Cache] added support for phpredis 4 `compression` and `tcp_keepalive` options
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] serialize objects using native arrays when possible
| 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 allows leveraging OPCache shared memory when storing objects in `Php*` pool storages (as done by default for all system caches). This improves performance a bit further when loading e.g. annotations, etc. (bench coming);
Instead of using native php serialization, this uses a marshaller that represents objects in plain static arrays. Unmarshalling these arrays is faster than unserializing the corresponding PHP strings (because it works with copy-on-write, while unserialize cannot.)
php-serialization is still a possible format because we have to use it when serializing structures with internal references or with objects implementing `Serializable`. The best serialization format is selected automatically so this is completely seamless.
ping @palex-fpt since you gave me the push to work on this, and are pursuing a similar goal in #27484. I'd be thrilled to get some benchmarks on your scenarios.
Commits
-------
866420e2eb [Cache] serialize objects using native arrays when possible
This PR was merged into the 4.2-dev branch.
Discussion
----------
[Cache] Improve perf of array-based pools
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | no
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | -
| License | MIT
| Doc PR | -
- skip key validation when key is already known
- remove overhead in `ArrayCache::get()` via inlining
- don't store simple values in serialized format when not needed, preserving COW
~~Needs #27565 to be green.~~
Commits
-------
92a2d4754f [Cache] Improve perf of array-based pools
* 4.1:
remove HHVM code
[VarDumper] Fix dumping ArrayObject and ArrayIterator instances
[ProxyManagerBridge] Fixed support of private services
[Cache] Fix typo in comment.
[FrameworkBundle] give access to non-shared services when using test.service_container
Fix bad method call with guard authentication + session migration
Avoid calling eval when there is no script embedded in the toolbar
* 3.4:
[VarDumper] Fix dumping ArrayObject and ArrayIterator instances
[ProxyManagerBridge] Fixed support of private services
[Cache] Fix typo in comment.
Fix bad method call with guard authentication + session migration
