* Chekote/xmlfileloader_overload_refactor:
Refactored the processing of each individual node into it's own method, enabling easier overloading of behavior for Bundles such as FriendsOfSymfony/RestBundle
This improves, for example, the exception one would receive if they tried to import a resource from a bundle that doesn't exist.
Previously, the deep "bundle is not activated" exception would be thrown. That has value, however there is no indication of where
the exception is actually occurring.
In this new implementation, we throw an exception that explains exactly which resource, and from which source resource, cannot be
loaded. The deeper exception is still thrown as a nested exception.
Two caveats:
* The `HttpKernel::varToString` method was replicated
* This introduces a new `Exception` class, which allows us to prevent lot's of exceptions from nesting into each other in the case
that some deeply imported resource cannot be imported (each upstream import that fails doesn't add its own exception).
* add unit tests
* introduce filename filter to DirectoryResource (to restrict change monitoring to a subset of files)
* modify AnnotationDirectoryLoader.php to use filename filter set on Controller\.php$
in your routing configuration, only existing files in this directory
get tracked for changes in this directory. So if you add a new
controller file in this directory you'd have to manually clear the
cache since the new file gets ignored.
This patch adds a DirectoryResource for this case which tracks all
changes in the given directory (and files and directories contained
within).
XML/YAML loaders assume imported resources are files before attempting to resolve their loader. This is problematic for loaders such as Assetic, which does not use a file as its resource. Furthermore, the previous consecutive calls to both locate() and getAbsolutePath() were redundant. File location can safely be delayed until FileLoader::import(), and we can let that throw an exception if the file is not found.
This reverts commit f53080860a.
Revert "[Router] config fixes"
This reverts commit 51beecc6f2.
Revert "moved duplicated files to a new Config component"
This reverts commit a8ec9b27f0.
Before I explain the changes, let's talk about the current state.
Before this patch, the registerBundleDirs() method returned an ordered (for
resource overloading) list of namespace prefixes and the path to their
location. Here are some problems with this approach:
* The paths set by this method and the paths configured for the autoloader
can be disconnected (leading to unexpected behaviors);
* A bundle outside these paths worked, but unexpected behavior can occur;
* Choosing a bundle namespace was limited to the registered namespace
prefixes, and their number should stay low enough (for performance reasons)
-- moreover the current Bundle\ and Application\ top namespaces does not
respect the standard rules for namespaces (first segment should be the
vendor name);
* Developers must understand the concept of "namespace prefixes" to
understand the overloading mechanism, which is one more thing to learn,
which is Symfony specific;
* Each time you want to get a resource that can be overloaded (a template for
instance), Symfony would have tried all namespace prefixes one after the
other until if finds a matching file. But that can be computed in advance
to reduce the overhead.
Another topic which was not really well addressed is how you can reference a
file/resource from a bundle (and take into account the possibility of
overloading). For instance, in the routing, you can import a file from a
bundle like this:
<import resource="FrameworkBundle/Resources/config/internal.xml" />
Again, this works only because we have a limited number of possible namespace
prefixes.
This patch addresses these problems and some more.
First, the registerBundleDirs() method has been removed. It means that you are
now free to use any namespace for your bundles. No need to have specific
prefixes anymore. You are also free to store them anywhere, in as many
directories as you want. You just need to be sure that they are autoloaded
correctly.
The bundle "name" is now always the short name of the bundle class (like
FrameworkBundle or SensioCasBundle). As the best practice is to prefix the
bundle name with the vendor name, it's up to the vendor to ensure that each
bundle name is unique. I insist that a bundle name must be unique. This was
the opposite before as two bundles with the same name was how Symfony2 found
inheritance.
A new getParent() method has been added to BundleInterface. It returns the
bundle name that the bundle overrides (this is optional of course). That way,
there is no ordering problem anymore as the inheritance tree is explicitely
defined by the bundle themselves.
So, with this system, we can easily have an inheritance tree like the
following:
FooBundle < MyFooBundle < MyCustomFooBundle
MyCustomFooBundle returns MyFooBundle for the getParent() method, and
MyFooBundle returns FooBundle.
If two bundles override the same bundle, an exception is thrown.
Based on the bundle name, you can now reference any resource with this
notation:
@FooBundle/Resources/config/routing.xml
@FooBundle/Controller/FooController.php
This notation is the input of the Kernel::locateResource() method, which
returns the location of the file (and of course it takes into account
overloading).
So, in the routing, you can now use the following:
<import resource="@FrameworkBundle/Resources/config/internal.xml" />
The template loading mechanism also use this method under the hood.
As a bonus, all the code that converts from internal notations to file names
(controller names: ControllerNameParser, template names: TemplateNameParser,
resource paths, ...) is now contained in several well-defined classes. The
same goes for the code that look for templates (TemplateLocator), routing
files (FileLocator), ...
As a side note, it is really easy to also support multiple-inheritance for a
bundle (for instance if a bundle returns an array of bundle names it extends).
However, this is not implemented in this patch as I'm not sure we want to
support that.
How to upgrade:
* Each bundle must now implement two new mandatory methods: getPath() and
getNamespace(), and optionally the getParent() method if the bundle extends
another one. Here is a common implementation for these methods:
/**
* {@inheritdoc}
*/
public function getParent()
{
return 'MyFrameworkBundle';
}
/**
* {@inheritdoc}
*/
public function getNamespace()
{
return __NAMESPACE__;
}
/**
* {@inheritdoc}
*/
public function getPath()
{
return strtr(__DIR__, '\\', '/');
}
* The registerBundleDirs() can be removed from your Kernel class;
* If your code relies on getBundleDirs() or the kernel.bundle_dirs parameter,
it should be upgraded to use the new interface (see Doctrine commands for
many example of such a change);
* When referencing a bundle, you must now always use its name (no more \ or /
in bundle names) -- this transition was already done for most things
before, and now applies to the routing as well;
* Imports in routing files must be changed:
Before: <import resource="Sensio/CasBundle/Resources/config/internal.xml" />
After: <import resource="@SensioCasBundle/Resources/config/internal.xml" />
Currently, ambiguities only arise for PHP files, as PhpFileLoader and AnnotationFileLoader would both claim support. Future conflicts may occur if the XML, YAML, or PHP loaders were to receive Directory and Glob loaders (as annotations have).
Since the "type" parameter is optional, loader resolution will default to awarding resolution to the first loader to claim support. A previous hack in PhpFileLoader to avoid an AnnotationFileLoader conflict was removed, so that should be the only lost backwards compatibility with this patch. Unit tests were also created for the various loader classes, although only the supports() method is being tested.
This implementation was proposed on the symfony-dev mailing list in response to Fabien's RFC for custom loader notation: http://groups.google.com/group/symfony-devs/browse_thread/thread/3104c1a9e45799d2/20fbe393c1afe088
When an object has a "main" many relation with related "things" (objects,
parameters, ...), the method names are normalized:
* get()
* set()
* all()
* replace()
* remove()
* clear()
* isEmpty()
* add()
* register()
* count()
* keys()
The classes below follow this method naming convention:
* BrowserKit\CookieJar -> Cookie
* BrowserKit\History -> Request
* Console\Application -> Command
* Console\Application\Helper\HelperSet -> HelperInterface
* DependencyInjection\Container -> services
* DependencyInjection\ContainerBuilder -> services
* DependencyInjection\ParameterBag\ParameterBag -> parameters
* DependencyInjection\ParameterBag\FrozenParameterBag -> parameters
* DomCrawler\Form -> FormField
* EventDispatcher\Event -> parameters
* Form\FieldGroup -> Field
* HttpFoundation\HeaderBag -> headers
* HttpFoundation\ParameterBag -> parameters
* HttpFoundation\Session -> attributes
* HttpKernel\Profiler\Profiler -> DataCollectorInterface
* Routing\RouteCollection -> Route
* Security\Authentication\AuthenticationProviderManager -> AuthenticationProviderInterface
* Templating\Engine -> HelperInterface
* Translation\MessageCatalogue -> messages
The usage of these methods are only allowed when it is clear that there is a
main relation:
* a CookieJar has many Cookies;
* a Container has many services and many parameters (as services is the main
relation, we use the naming convention for this relation);
* a Console Input has many arguments and many options. There is no "main"
relation, and so the naming convention does not apply.
For many relations where the convention does not apply, the following methods
must be used instead (where XXX is the name of the related thing):
* get() -> getXXX()
* set() -> setXXX()
* all() -> getXXXs()
* replace() -> setXXXs()
* remove() -> removeXXX()
* clear() -> clearXXX()
* isEmpty() -> isEmptyXXX()
* add() -> addXXX()
* register() -> registerXXX()
* count() -> countXXX()
* keys()
This is not really a bug or a big issue, but if you need to temporarily empty a routes node, Symfony shouldn't care - it's not an actual validation failure (a routes node still makes sense without an route or import children).
