Overview

Other MyFaces Extensions

Community

Development

Sponsorship

Unknown macro: {iframe}

Your browser does not support iframes

Skip to end of metadata
Go to start of metadata

Table of Contents

Intro

See Pre-required readings

Further readings:

Usage for Users

ProjectStage

Project stages allow to use implementations depending on the current environment. E.g. you can implement a bean which creates sample-data for unit tests which gets activated only in case of project-stage UnitTest.

Besides custom project-stages it's possible to use the following pre-defined project-stages:

  • UnitTest
  • Development
  • SystemTest
  • SystemTest
  • IntegrationTest
  • Staging
  • Production

The core provides a pluggable and type-safe approach for using project stages in a project (it's also used within the framework). Furthermore, @ProjectStageActivated allows to use e.g. implementations annotated with javax.enterprise.inject.Alternative for specific project-stages. Besides the out-of-the-box project-stages it's possible to implement custom but type-safe project-stages which will be exposed by CODI.

Resolving and using the Project-Stage

Activation

@ProjectStageActivated

This annotation allows to activate beans for a special project-stage. It's possible to use one of the out-of-the-box project-stages or a custom typesafe project-stage.

Alternative implementation activated for Project-Stage UnitTest
Alternative implementation activated for Project-Stages UnitTest and Development

To configure a project-stage you can use the key: org.apache.myfaces.extensions.cdi.ProjectStage and configure it for your environment (see the out-of-the-box environment-config options). Static configuration files like web.xml and property files aren't supported by default because in our opinion it's a wrong and quite risky place to configure it independent of the environment - however, with a custom ConfiguredValueResolver it's possible to resolve the project-stage configuration from any kind of configuration-source.
If there is no CODI project-stage configured and it's a JSF application, CODI will re-use the project-stage configured for JSF (that's possible because JSF project-stages are a logical subset of CODI project-stages and project stages which are identical also use the same name and will be mapped automatically).

@ExpressionActivated

This annotation allows to activate beans based on expressions. Out-of-the-box simple conditions are supported. The values will be resolved from the environment via the default resolvers (see out-of-the-box environment-config options) or via a custom ConfiguredValueResolver which allows to use any type of configuration-format.

Alternative implementation activated based on a configured value

'*' allows to ensure that there is a configured value. So the 2nd example ensures that there is a value for 'db' but it isn't 'prod'.

(Custom) ExpressionInterpreter

Per default a simple syntax for key/value based configs is supported. However, it's possible to implement a custom interpreter for custom expressions.

A custom interpreter for custom expressions

Config

The core of CODI provides base implementations which are used to provide a testable, pluggable and type-safe config. If it is possible this CDI based config approach is used to configure the framework (esp. the SPI). This approach decouples the logical config entries from the config-source. That means it's possible to switch e.g. between Java Config (values are directly encoded in config classes), property files, db table/s, xml (e.g. web.xml) and every other format which can be used for a config.

Out-of-the-box CODI uses typesafe configuration. Furthermore, there is an alternative configuration module which allows to resolve configure values from myfaces-extcdi.properties or via a custom ConfiguredValueResolver from any other configuration format/source.

Hint

Since the config is based on CDI mechanisms, it isn't possible to use it for all configurations. There are some very basic configurations (e.g. deactivation of specific CODI features), which have to be accessible before CDI is fully initialized. Furthermore, the new versions of Weld (at least until the version which implements CDI v1.1) requires a special bundled version to use a custom configuration.

CodiConfig

All modules which have CODI-Core as a dependency use the CodiConfig interface as marker interface for their config-classes. So it's easy to find all configurations. All methods annotated with ConfigEntry provide a config-value which is used by CODI. During bootstrapping all values are logged (it's possible to deactivate this behaviour by config see CodiCoreConfig#isConfigurationLoggingEnabled). If you provide custom values and you face any issue, please also post the configuration to the mailing-list.

CodiCoreConfig

This config contains the basic configuration for the core as well as common config entries for multiple modules.

Environment-Config Options

In some cases it isn't possible to use CDI based configuration. Esp. for parts which have to be active before the CDI container gets bootstrapped.
For such cases CODI uses an extensible configuration approach. By default it supports configuration via:

  • ServiceLoader (if classes have to be configured which implement an interface or extend an abstract class)
  • System properties
  • JNDI
  • Resource Bundle (properties file with the name myfaces-extcdi.properties btw. /META-INF/myfaces-extcdi.properties)

However, with a custom ConfiguredValueResolver it's possible to resolve configured values from any kind of configuration-source.

If the custom implementation is annotated with @Advanced it's possible to wrap the default implementation for delegating to it. The only requirement is a field called default + the name of the artifact - e.g. defaultViewConfigExtractor . Since such artifacts are used during the bootstrapping process CDI, it isn't possible to use std. CDI based dependency injection.

Logging

To avoid external dependencies, CODI uses the JDK Logger. However, CDI often requires serializable beans and JDK loggers aren't serializable - therefore CODI provides a serializable wrapper for the JDK Logger.

Injectable JDK logger

By default the fully qualified class name of the target class which uses the injected logger, will be used to create the logger. As an alternative it's possible to use the LoggerDetails qualifier to provide e.g. a name for the logger.

Injectable JDK logger with a custom name

Provider

BeanManagerProvider

Resolving the Bean-Manager

In some cases (e.g. in custom CDI extensions or classes which aren't managed by CDI), it isn't possible to inject the BeanManager. Therefore, it's possible to use BeanManagerProvider.getInstance().getBeanManager().

Startup

CODI provides multiple hooks for the startup. Usually it's enough to observe the StartupEvent fired by CODI during the startup-process as soon as the target environment is up and running. In case of JSF this event is fired lazily. If you need to execute custom logic before CODI gets active, you should have a look at the dev guide (see StartupEventBroadcaster).

Observing the startup event

Type-safe Resource-bundles (since CODI v1.0.1)

The message-module of CODI allows easy, pluggable and therefore advanced message-handling. However, besides messages it's sometimes essential to use resource-bundles in a type-safe manner (e.g. in case of custom configs which don't need be customizable).

Injecting and using a Resource-bundle

@Bundle allows to inject a ResourceBundle. This interface is provided by CODI and is a simpler but injectable version of the std. ResourceBundle.

Injecting and using a resource-bundle

By default a resource-bundle class/interface is mapped to the corresponding bundle-file via naming convention. A class/interface can be annotated with @Bundle optionally to find it easier via searching for the annotation or to changing the name and/or package of the corresponding bundle-file.

Implementing a resource-bundle

Injecting a Resource-bundle value directly

Instead of injecting the resource-bundle and resolving a value by key, it's possible to inject the value directly. That's e.g. useful for configs, because in such cases you are interested in few very specific values.

Injecting a resource-bundle value

Tools

DefaultAnnotation

For creating instances of Annotations, you can use the literal trick. A custom implementation allows to provide custom values (see e.g. the NamedLiteral which is used by CODI internally). If you are fine with the default values of an annotation, you can use DefaultAnnotation to create an annotation for a given type, instead of a custom literal implementation.

Create instances of Annotations

Misc

@Advanced

This annotation can be used as marker annotation for artifacts which aren't managed by CDI. If such arifacts are supported by CODI (the documentation of every module shows such artifacts) and annotated with this annotation, it's possible to use dependency injection for fields. Furthermore, this annotation can be used as CDI qualifier.

@Enhanced

This annotation isn't used by CODI modules. It's a common annotation which can be used as marker annotation or qualifier by CODI add-ons. So they don't have to introduce their own marker annotation.

BeanNames

This interface is a marker interface for all interfaces which contain bean names (used by @javax.inject.Named within CODI). So it's very easy to find beans which can be used in EL expressions.

CoreModuleBeanNames

This interface specifies all beans of the core which can be resolved by name.

@InvocationOrder

Artifacts which support this annotation can be sorted. The default order is 1000. Artifacts with a lower ordinal will be called before artifacts with a higher ordinal. If there is an artifact which is supported by CODI but doesn't provide an explicit order, it will be moved at the end of the list.

Base contracts

The following artifacts define base contracts which can be used by CODI modules or any other custom modules which would like to build on the well known contracts provided by CODI.
The information provided by this section is quite general because this part of the Core just specifies a quite abstract contract which is independent of a concrete technology.

Please have a look at the specific modules which use those contracts (esp. the JSF module), if you need more concrete information.

Config/View

This part of the Core provides the basic concepts which allow to provide a type-safe config for views. CODI currently uses it in the JSF module for typesafe navigation as well as multiple inheritance for type-safe page-configs.
ViewConfig as well as DefaultErrorView are the foundations for the concept and will be explained in the documentation of the JSF module. However, it's possible to use it for other UI technologies as well.

@View

This annotation can be used as interceptor to restrict calls to methods to one or more views. Modules are also allowed to use it as marker annotation to link a bean to a view by annotating the bean with it and pointing to the view(s). In this case a module is allowed to store the information internally and remove the interceptor dynamically to avoid performance impacts.

ViewMetaData

This annotation allows to provide custom meta-data. Just annotate a custom annotation with it. A module like the JSF module has to provide a resolver to query it. A query might return multiple results of the same type. If it doesn't make sense to have multiple results, you can use @ViewMetaData(override=true).

Custom Meta-data for View-Configs

The JSF module of CODI allows to resolve this meta-data via the ViewConfigResolver. Since this resolver is aware of JSF specific concepts, it isn't provided by the Core. Please have a look at the documentation of the JSF module to get further information about it.

Navigation

ViewNavigationHandler

This navigation handler defines how to navigate with view-configs. Currently it's used by all JSF modules to allow implicit navigation based on the view-configs.

Observing the navigation

This example is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Since it's a generic concept, it's also possible to provide an implementations for other view-technologies.

PreViewConfigNavigateEvent

This event gets triggered if a navigation is going to happen from and to a page represented by a view-config. It also allows to redefine the navigation target.

Observing the navigation

In this example the navigation is observed and in a special case the navigation target is changed to the configured default error view. At the end a message gets created and added to the current MessageHandler. In case of a JSF application it gets added to the FacesContext.

This example is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

Scope/Conversation

All interfaces and annotations which are independent from the concrete UI technology are provided by the core. E.g. the concepts of CODI conversations aren't bound to JSF. So the core defines the basic API for CODI conversations. Instead of providing a framework adapter (like MyFaces Orchestra does), CODI hosts the specific implementation in the corresponding module. Currently the UI modules of CODI are just for JSF. However, it would be possible to provide a module e.g. for (plain) servlets. Such a module can use the APIs provided by the core (which are independent of JSF and its concepts).

@ConversationScoped

This isn't the std. CDI annotation!

With this annotation it's possible to scope beans in parallel or grouped conversations which are way more powerful that the std. conversation of CDI 1.

This scope is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

@ViewAccessScoped

With this annotation it's possible to save beans until the first request of a new view doesn't access it. It's a very convenient scope e.g. for wizard which can't be interrupted.

This scope is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

@WindowScoped

This annotation is comparable to a session per window. Compared to the scopes mentioned above, all beans will be removed at the same point in time.

This scope is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

WindowContext

This context contains all CODI scopes which are bound to a window and allows to control them in a fine-grained manner. Furthermore, it contains information about the current window like the window-id.

This interface is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

Conversation

This interface allows to close the associated conversation immediately. Furthermore, it's possible to restart the conversation if it is known that it's going to started again soon (the only difference to closing the conversation is a better performance).

This interface is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

@ConversationGroup

This annotation allows to group conversations with a type-safe mechanism.

This interface is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

@ConversationRequired

This annotation allows to prevent a navigation to a view which require an existing conversation.

This interface is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

@CloseConversationGroup

This method-interceptor allows to close the conversation(-group) of the current bean or an explicitly specified group after the invocation of the method or in case of a specified exception.

This interface is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

Scope/Conversation/Config

ConversationConfig

The conversation config allows to specify e.g. the time-out for grouped conversations and to enable the optional events.

This config is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

WindowContextConfig

The window config allows to specify e.g. the time-out for a window which includes all CODI scopes and to adjust the basic window/-id handling.

This config is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Scope/Conversation/Event

The following events are optional and have to be activated via the configs mentioned above.

  • AccessBeanEvent
  • CloseConversationEvent
  • CloseWindowContextEvent
  • CreateWindowContextEvent
  • RestartConversationEvent
  • ScopeBeanEvent
  • StartConversationEvent
  • UnscopeBeanEvent

Security

CODI provides some basic hooks to integrate a custom security concept. It isn't related to an existing framework. CODI just allows to integrate such frameworks based on the ViewConfig or via interceptors.

AccessDecisionVoter

This interface is (besides the Secured annotation) the most important part of the concept. Both artifact types are also the only required parts.
This interface is independent of JSF. However, currently you need the JSF 1.x or 2.x module to use an out-of-the-box implementation of it. Further, details and examples are available in the documentation of the CODI-JSF-module.

Since it's a generic concept, it's also possible to provide an implementation for other view-technologies.

The AbstractAccessDecisionVoter allows an easier implementation.

@Secured

Simple usage of @Secured for ViewConfigs

In case of a violation CODI will use the DefaultErrorView as navigation target (if configured).

Simple usage of @Secured for ViewConfigs with a special error page

In this case the page represented by Login.class with be used instead of the DefaultErrorView.

As an alternative it's possible to use the annotation as interceptor for beans.

Alternative usage of @Secured as interceptor

SecurityViolation

In case of a detected violation a SecurityViolation has to be added to the result returned by the AccessDecisionVoter.

Simple example for creating a SecurityViolation

The rest is done by CODI. Please note that there is a natural overhead if the @Secured annotation is used as interceptor. In combination with the JSF module, we recommend to us it for the ViewConfig instead of beans because the performance overhead is minimal compared to an interceptor.

@Secured and Stereotypes (since CODI v1.0.4)

If there are multiple AccessDecisionVoter and maybe in different constellations, it's easier to provide an expressive CDI stereotypes for it. Later on that also allows to change the behaviour in a central place.

Stereotype support of @Secured

Furthermore, it's possible to provide custom meta-data easily.

Stereotype of @Secured with custom meta-data

BeanCreationDecisionVoter

This feature is optional and has to be activated via CodiCoreConfig#isInvalidBeanCreationEventEnabled. It allows to check the permission before a bean gets created for a CODI scope.
The abstract class AbstractBeanCreationDecisionVoter allows an easier implementation. An InvalidBeanCreationEvent will be fired for every detected violation and it's possible to prevent the creation of an exception (see #setThrowSecurityViolation) because a context doesn't allow to handle such exceptions in a nice way. Internally CODI uses this concept for @ConversationRequired, however, it's possible to use it also for custom concepts.

AccessDeniedException

This exception will be thrown in case of a security violation and contains the reason(s) of the violation(s) (a set of SecurityViolation )