Quarkus - Writing Your Own Extension

Quarkus extensions add a new developer focused behavior to the core offering, and consist of two distinct parts, buildtime augmentation and runtime container. The augmentation part is responsible for all metadata processing, such as reading annotations, XML descriptors etc. The output of this augmentation phase is recorded bytecode which is responsible for directly instantiating the relevant runtime services.

This means that metadata is only processed once at build time, which both saves on startup time, and also on memoryusage as the classes etc that are used for processing are not loaded (or even present) in the runtime JVM.

1. Extension philosophy

This section is a work in progress and gathers the philosophy under which extensions should be designed and written.

1.1. Why an extension framework

Quarkus’s mission is to transform your entire application including the libraries it uses, into an artifact that uses significantly less resources than traditional approaches. These can then be used to build native applications using GraalVM.To do this you need to analyze and understand the full "closed world" of the application.Without the full and complete context, the best that can be achieved is partial and limited generic support.By using the Quarkus extension approach, we can bring Java applications in line with memory footprint constrained environments like Kubernetes or cloud platforms.

The Quarkus extension framework results in significantly improved resource utilization even when GraalVM is not used (e.g. in HotSpot).Let’s list the actions an extension performs:

  • Gather build time metadata and generate code
  • This part has nothing to do with GraalVM, it is how Quarkus starts frameworks “at build time”

  • The extension framework facilitates reading metadata, scanning classes as well as generating classes as needed

  • A small part of the extension work is executed at runtime via the generated classes, while the bulk of the work is done at build time (called deployment time)

  • Enforce opinionated and sensible defaults based on the close world view of the application (e.g. an application with no @Entity does not need to start Hibernate ORM)

  • An extension hosts Substrate VM code substitution so that libraries can run on GraalVM

  • Most changes are pushed upstream to help the underlying library run on GraalVM

  • Not all changes can be pushed upstream, extensions host Substrate VM substitutions - which is a form of code patching - so that libraries can run

  • Host Substrate VM code substitution to help dead code elimination based on the application needs
  • This is application dependant and cannot really be shared in the library itself

  • For example, Quarkus optimizes the Hibernate code because it knows it only needs a specific connection pool and cache provider

  • Send metadata to GraalVM for example classes in need of reflection
  • This information is not static per library (e.g. Hibernate) but the framework has the semantic knowledge and knows which classes need to have reflection (for example @Entity classes)

1.2. Favor build time work over runtime work

As much as possible favor doing work at build time (deployment part of the extension) as opposed to let the framework do work at startup time (runtime).The more is done there, the smaller Quarkus applications using that extension will be and the faster they will load.

1.3. How to expose configuration

Quarkus simplifies the most common usages.This means that its defaults might be different than the library it integrates.

To make the simple experience easiest, unify the configuration in application.properties via MicroProfile Config.Avoid library specific configuration files, or at least make them optional: e.g. persistence.xml for Hibernate ORM is optional.

Extensions should see the configuration holistically as a Quarkus application instead of focusing on the library experience.For example quarkus.database.url and friends are shared between extensions as defining a database access is a shared task (instead of a hibernate. property for example).The most useful configuration options should be exposed as quarkus.[extension]. instead of the natural namespace of the library.Less common properties can live in the library namespace.

To fully enable the close world assumptions that Quarkus can optimize best, it is better to consider configuration options as build time settled vs overridable at runtime.Of course properties like host, port, password should be overridable at runtime.But many properties like enable caching or setting the JDBC driver can safely require a rebuild of the application.

1.4. Expose your components via CDI

Since CDI is the central programming model when it comes to component composition, frameworks should expose producers that are easily consumable by user applications.For example, Hibernate ORM exposes EntityManagerFactory and EntityManager beans, the connection pool exposes DataSource beans etc.An extension must generate these bean definitions at build time.

A very useful pattern of creating such beans but also giving application code the ability to easily override some of the beans with custom implementations, is to usethe @DefaultBean that Quarkus provides. This is best explained with an example.

1.4.1. Example CDI configuration using @DefaultBean

Let us assume that the Quarkus extension needs to provide a Tracer bean which application code is meant to inject into its own beans.

  1. @Dependent
  2. public class TracerConfiguration {
  3. @Produces
  4. public Tracer tracer(Reporter reporter, Configuration configuration) {
  5. return new Tracer(reporter, configuration);
  6. }
  7. @Produces
  8. @DefaultBean
  9. public Configuration configuration() {
  10. // create a Configuration
  11. }
  12. @Produces
  13. @DefaultBean
  14. public Reporter reporter(){
  15. // create a Reporter
  16. }
  17. }

If for example application code wants to use Tracer, but also needs to use a custom Reporter bean, such a requirement could easily be done using something like:

  1. @Dependent
  2. public class CustomTracerConfiguration {
  3. @Produces
  4. public Reporter reporter(){
  5. // create a custom Reporter
  6. }
  7. }

1.4.2. How to Override a Bean Defined by a Library/Quarkus Extension that doesn’t use @DefaultBean

Although @DefaultBean is the recommended approach, it is also possible for application code to override beans provided by an extension by marking beans as a CDI @Alternative and including @Priority annotation.Let’s show a simple example.Suppose we work on an imaginary "quarkus-parser" extension and we have a default bean implementation:

  1. @Dependent
  2. class Parser {
  3. String[] parse(String expression) {
  4. return expression.split("::");
  5. }
  6. }

And our extension also consumes this parser:

  1. @ApplicationScoped
  2. class ParserService {
  3. @Inject
  4. Parser parser;
  5. //...
  6. }

Now, if a user or even some other extension needs to override the default implementation of the Parser the simplest solution is to use CDI @Alternative + @Priority:

  1. @Alternative (1)
  2. @Priority(1) (2)
  3. @Singleton
  4. class MyParser extends Parser {
  5. String[] parse(String expression) {
  6. // my super impl...
  7. }
  8. }
1MyParser is an alternative bean.
2Enables the alternative. The priority could be any number to override the default bean but if there are multiple alternatives the highest priority wins.
CDI alternatives are only considered during injection and type-safe resolution. For example the default implementation would still receive observer notifications.

1.5. Some types of extensions

There exist multiple stereotypes of extension, let’s list a few.

  • Bare library running
  • This is the less sophisticated extension.It consists of a set of patches to make sure a library runs on GraalVM.If possible, contribute these patches upstream, not in extensions.Second best is to write Substrate VM substitutions, which are patches applied during native image compilation.

  • Get a framework running

  • A framework at runtime typically reads configuration, scan the classpath and classes for metadata (annotations, getters etc), build a metamodel on top of which it runs, find options via the service loader pattern, prepare invocation calls (reflection), proxy interfaces, etc.These operations should be done at build time and the metamodel be passed to the recorder DSL that will generate classes that will be executed at runtime and boot the framework.

  • Get a CDI portable extension running

  • The CDI portable extension model is very flexible.Too flexible to benefit from the build time boot promoted by Quarkus.Most extension we have seen do not make use of these extreme flexibilities capabilities.The way to port a CDI extension to Quarkus is to rewrite it as a Quarkus extension which will define the various beans at build time (deployment time in extension parley).

2. Technical aspect

2.1. Three Phases of Bootstrap and Quarkus Philosophy

There are three distinct bootstrap phases of a Quarkus app:

  • Augmentation
  • This is the first phase, and is done by the Build Step Processors. These processors have access to Jandex annotationinformation and can parse any descriptors and read annotations, but should not attempt to load any application classes. The output of thesebuild steps is some recorded bytecode, using an extension of the ObjectWeb ASM project called Gizmo(ext/gizmo), that is used to actually bootstrap the application at runtime. Depending on the io.quarkus.deployment.annotations.ExecutionTime value of the @io.quarkus.deployment.annotations.Record annotation associated with the build step,the step may be run in a different JVM based on the following two modes.

  • Static Init

  • If bytecode is recorded with @Record(STATIC_INIT) then it will be executed from a static init method on the mainclass. For a native executable build, this code is executed in a normal JVM as part of the native buildprocess, and any retained objects that are produced in this stage will be directly serialized into the native executable via an image mapped file.This means that if a framework can boot in this phase then it will have its booted state directly written to theimage, and so the boot code does not need to be executed when the image is started.

There are some restrictions on what can be done in this stage as the Substrate VM disallows some objects in the native executable. For example you should not attempt to listen on a port or start threads in this phase.

In non-native pure JVM mode, there is no real difference between Static and Runtime Init, except that Static Init is always executed first. This mode benefits from the same build phase augmentation as native mode as the descriptor parsing and annotation scanning are doneat build time and any associated class/framework dependencies can be removed from the build output jar. In servers likeWildFly, deployment related classes such as XML parsers hang around for the life of the application, using up valuablememory. Quarkus aims to eliminate this, so that the only classes loaded at runtime are actually used at runtime.

As an example, the only reason that a Quarkus application should load an XML parser is if the user is using XML in theirapplication. Any XML parsing of configuration should be done in the Augmentation phase.

  • Runtime Init
  • If bytecode is recorded with @Record(RUNTIME_INIT) then it is executed from the application’s main method. This codewill be run on native executable boot. In general as little code as possible should be executed in this phase, and shouldbe restricted to code that needs to open ports etc.

Pushing as much as possible into the @Record(STATIC_INIT) phase allows for two different optimizations:

  • In both native executable and pure JVM mode this allows the app to start as fast as possible since processing was done during build time. This also minimizes the classes/native code needed in the application to pure runtime related behaviors.

  • Another benefit with native executable mode is that Substrate can more easily eliminate features that are not used. If features are directly initialized via bytecode, Substrate can detect that a method is never called and eliminatethat method. If config is read at runtime, Substrate cannot reason about the contents of the config and so needs to keep all features in case they are required.

2.2. Maven setup

Your extension project should be setup as a multi-module project with two submodules:

  • A deployment time submodule that handles the build time processing and bytecode recording.

  • A runtime submodule that contains the runtime behavior that will provide the extension behavior in the native executable or runtime JVM.

You may want to use the create-extension mojo of io.quarkus:quarkus-maven-plugin to create these Maven modules - see the next section.

Your runtime artifact should depend on io.quarkus:quarkus-core, and possibly the runtime artifacts of other Quarkusmodules if you want to use functionality provided by them.You will also need to include the io.quarkus:quarkus-bootstrap-maven-plugin to generate the Quarkus extension descriptor included into the runtime artifact, if you are using the Quarkus parent pom it will automatically inherit the correct configuration.Futhermore, you’ll need to configure the maven-compiler-plugin to detect the quarkus-extension-processor annotation processor.

By convention the deployment time artifact has the -deployment suffix, and the runtime artifacthas no suffix (and is what the end user adds to their project).
  1. <dependencies>
  2. <dependency>
  3. <groupId>io.quarkus</groupId>
  4. <artifactId>quarkus-core</artifactId>
  5. </dependency>
  6. </dependencies>
  7. <build>
  8. <plugins>
  9. <plugin>
  10. <groupId>io.quarkus</groupId>
  11. <artifactId>quarkus-bootstrap-maven-plugin</artifactId>
  12. <!-- Executions configuration can be inherited from quarkus-build-parent -->
  13. <executions>
  14. <execution>
  15. <goals>
  16. <goal>extension-descriptor</goal>
  17. </goals>
  18. <configuration>
  19. <deployment>${project.groupId}:${project.artifactId}-deployment:${project.version}</deployment>
  20. </configuration>
  21. </execution>
  22. </executions>
  23. </plugin>
  24. <plugin>
  25. <groupId>org.apache.maven.plugins</groupId>
  26. <artifactId>maven-compiler-plugin</artifactId>
  27. <configuration>
  28. <annotationProcessorPaths>
  29. <path>
  30. <groupId>io.quarkus</groupId>
  31. <artifactId>quarkus-extension-processor</artifactId>
  32. </path>
  33. </annotationProcessorPaths>
  34. </configuration>
  35. </plugin>
  36. </plugins>
  37. </build>
The above maven-compiler-plugin configuration requires version 3.5+.
Under no circumstances can the runtime module depend on a deployment artifact. This would resultin pulling all the deployment time code into runtime scope, which defeats the purpose of having the split.

Your deployment time module should depend on io.quarkus:quarkus-core-deployment, your runtime artifact,and possibly the deployment artifacts of other Quarkus modules if you want to use functionality provided by them.You will also need to configure the maven-compiler-plugin to detect the quarkus-extension-processor annotation processor.

  1. <dependencies>
  2. <dependency>
  3. <groupId>io.quarkus</groupId>
  4. <artifactId>quarkus-core-deployment</artifactId>
  5. </dependency>
  6. </dependencies>
  7. <build>
  8. <plugins>
  9. <plugin>
  10. <groupId>org.apache.maven.plugins</groupId>
  11. <artifactId>maven-compiler-plugin</artifactId>
  12. <configuration>
  13. <annotationProcessorPaths>
  14. <path>
  15. <groupId>io.quarkus</groupId>
  16. <artifactId>quarkus-extension-processor</artifactId>
  17. </path>
  18. </annotationProcessorPaths>
  19. </configuration>
  20. </plugin>
  21. </plugins>
  22. </build>

2.2.1. Create new extension modules using Maven

The create-extension mojo of io.quarkus:quarkus-maven-plugin can be used to generate stubs of Maven modules needed for implementing a new Quarkus extension.

This mojo can be currently used only for adding extensions to an established source tree hosting multiple extensions in one subdirectory, such as Quarkus or Camel Quarkus. Creating extension projects from scratch is not supported yet.

As and example, let’s add a new extension called my-ext to the Quarkus source tree:

  1. git clone https://github.com/quarkusio/quarkus.git
  2. cd quarkus
  3. cd extensions
  4. mvn io.quarkus:quarkus-maven-plugin:1.0.0.CR1:create-extension -N \
  5. -Dquarkus.artifactIdBase=my-ext \
  6. -Dquarkus.artifactIdPrefix=quarkus- \
  7. -Dquarkus.nameBase="My Extension"

The above sequence of commands does the following:

  • Creates four new Maven modules:
  • quarkus-my-ext-parent in the extensions/my-ext directory

  • quarkus-my-ext in the extensions/my-ext/runtime directory

  • quarkus-my-ext-deployment in the extensions/my-ext/deployment directory; a basic MyExtProcessor class is generated in this module.

  • quarkus-my-ext-integration-test in the integration-tests/my-ext/deployment directory; an empty JAX-RS Resource class and two test classes (for JVM mode and native mode) are generated in this module.

  • Links these three modules where necessary:
  • quarkus-my-ext-parent is added to the <modules> of quarkus-extensions-parent

  • quarkus-my-ext is added to the <dependencyManagement> of the runtime BOM (Bill of Materials) bom/runtime/pom.xml

  • quarkus-my-ext-deployment is added to the <dependencyManagement> of the deployment BOM (Bill of Materials) bom/deployment/pom.xml

  • quarkus-my-ext-integration-test is added to the <modules> of quarkus-integration-tests-parent

A Maven build performed immediately after generating the modules should fail due to a fail() assertion in one of the test classes.

There is one step (specific to the Quarkus source tree) that you should do manually when creating a new extension:Add the extension coordinates to devtools/common/src/main/filtered/extensions.json.

Note that the parameters of the mojo that will be constant for all the extensions added to this source tree are configured in extensions/pom.xml so that they do not need to be passed on the command line each time a new extension is added:

  1. <plugin>
  2. <groupId>io.quarkus</groupId>
  3. <artifactId>quarkus-maven-plugin</artifactId>
  4. <version>${quarkus.version}</version>
  5. <inherited>false</inherited>
  6. <!-- Settings for stubbing new extensions via
  7. mvn quarkus:create-extension -N -Dquarkus.artifactIdBase=my-ext -Dquarkus.nameBase="My Extension"
  8. -->
  9. <configuration>
  10. <namePrefix xml:space="preserve">Quarkus - </namePrefix>
  11. <runtimeBomPath>../bom/runtime/pom.xml</runtimeBomPath>
  12. <deploymentBomPath>../bom/deployment/pom.xml</deploymentBomPath>
  13. <itestParentPath>../integration-tests/pom.xml</itestParentPath>
  14. </configuration>
  15. </plugin>
The nameBase parameter of the mojo is optional.If you do not specify it on the command line, the plugin will derive it from artifactIdBase by replacing dashes with spaces and uppercasing each token.So you may consider omitting explicit nameBase in some cases.

Please refer to CreateExtensionMojo JavaDoc for all the available options of the mojo.

2.3. Build Step Processors

Work is done at augmentation time by build steps which produce and consume build items. The build steps found inthe deployment modules that correspond to the extensions in the project build are automatically wired together and executedto produce the final build artifact(s).

2.3.1. Build steps

A build step is a method which is annotated with the @io.quarkus.deployment.annotations.BuildStep annotation. Eachbuild step may consume items that are produced by earlier stages, and produce itemsthat can be consumed by later stages. Build steps are normally only run when they produce a build item that isultimately consumed by another step.

Build steps are normally placed on plain classes within an extension’s deployment module. The classes are automaticallyinstantiated during the augment process and utilize injection.

2.3.2. Build items

Build items are concrete, final subclasses of the abstract io.quarkus.builder.item.BuildItem class. Each build item representssome unit of information that must be passed from one stage to another. The base BuildItem class may not itself be directlysubclassed; rather, there are abstract subclasses for each of the kinds of build item subclasses that may be created:simple, multi, and empty.

Think of build items as a way for different extensions to communicate with one another. For example, a build item can:

  • expose the fact that a database configuration exists

  • consume that database configuration (e.g. a connection pool extension or an ORM extension)

  • ask an extension to do work for another extension: e.g. an extension wanting to define a new CDI bean and asking the ArC extensionto do so

This is a very flexible mechanism.

BuildItem instances should be immutable, as the producer/consumer model does not allow for mutation to be correctlyordered. This is not enforced but failure to adhere to this rule can result in race conditions.
2.3.2.1. Simple build items

Simple build items are final classes which extend io.quarkus.builder.item.SimpleBuildItem. Simple build items may onlybe produced by one step in a given build; if multiple steps in a build declare that they produce the same simple build item,an error is raised. Any number of build steps may consume a simple build item. A build step which consumes a simplebuild item will always run after the build step which produced that item.

Example of a single build item

  1. /**
  2. * The build item which represents the Jandex index of the application,
  3. * and would normally be used by many build steps to find usages
  4. * of annotations.
  5. */
  6. public final class ApplicationIndexBuildItem extends SimpleBuildItem {
  7. private final Index index;
  8. public ApplicationIndexBuildItem(Index index) {
  9. this.index = index;
  10. }
  11. public Index getIndex() {
  12. return index;
  13. }
  14. }
2.3.2.2. Multi build items

Multiple or "multi" build items are final classes which extend io.quarkus.builder.item.MultiBuildItem. Any number ofmulti build items of a given class may be produced by any number of steps, but any steps which consume multi build itemswill only run after every step which can produce them has run.

Example of a multiple build item

  1. public final class ServiceProviderBuildItem extends MultiBuildItem {
  2. private final String serviceName;
  3. private final List<String> implementations;
  4. public ServiceProviderBuildItem(String serviceName, String... implementations) {
  5. this.serviceName = serviceName;
  6. // Make sure it's immutable
  7. this.implementations = Collections.unmodifiableList(
  8. Arrays.asList(
  9. implementations.clone()
  10. )
  11. );
  12. }
  13. public String getServiceName() {
  14. return serviceName;
  15. }
  16. public List<String> getImplementations() {
  17. return implementations;
  18. }
  19. }

Example of multiple build item usage

  1. /**
  2. * This build step produces a single multi build item that declares two
  3. * providers of one configuration-related service.
  4. */
  5. @BuildStep
  6. public ServiceProviderBuildItem registerOneService() {
  7. return new ServiceProviderBuildItem(
  8. Converter.class.getName(),
  9. MyFirstConfigConverterImpl.class.getName(),
  10. MySecondConfigConverterImpl.class.getName()
  11. );
  12. }
  13. /**
  14. * This build step produces several multi build items that declare multiple
  15. * providers of multiple configuration-related services.
  16. */
  17. @BuildStep
  18. public void registerSeveralSerivces(
  19. BuildProducer<ServiceProviderBuildItem> providerProducer
  20. ) {
  21. providerProducer.produce(new ServiceProviderBuildItem(
  22. Converter.class.getName(),
  23. MyThirdConfigConverterImpl.class.getName(),
  24. MyFourthConfigConverterImpl.class.getName()
  25. ));
  26. providerProducer.produce(new ServiceProviderBuildItem(
  27. ConfigSource.class.getName(),
  28. MyConfigSourceImpl.class.getName()
  29. ));
  30. }
  31. /**
  32. * This build step aggregates all the produced service providers
  33. * and outputs them as resources.
  34. */
  35. @BuildStep
  36. public void produceServiceFiles(
  37. List<ServiceProviderBuildItem> items,
  38. BuildProducer<GeneratedResourceBuildItem> resourceProducer
  39. ) throws IOException {
  40. // Aggregate all of the providers
  41. Map<String, Set<String>> map = new HashMap<>();
  42. for (ServiceProviderBuildItem item : items) {
  43. String serviceName = item.getName();
  44. for (String implName : item.getImplementations()) {
  45. map.computeIfAbsent(
  46. serviceName,
  47. (k, v) -> new LinkedHashSet<>()
  48. ).add(implName);
  49. }
  50. }
  51. // Now produce the resource(s) for the SPI files
  52. for (Map.Entry<String, Set<String>> entry : map.entrySet()) {
  53. String serviceName = entry.getKey();
  54. try (ByteArrayOutputStream os = new ByteArrayOutputStream()) {
  55. try (OutputStreamWriter w = new OutputStreamWriter(os, StandardCharsets.UTF_8)) {
  56. for (String implName : entry.getValue()) {
  57. w.write(implName);
  58. w.write(System.lineSeparator());
  59. }
  60. }
  61. w.flush();
  62. resourceProducer.produce(
  63. new GeneratedResourceBuildItem(
  64. "META-INF/services/" + serviceName,
  65. os.toByteArray()
  66. )
  67. );
  68. }
  69. }
  70. }
2.3.2.3. Empty build items

Empty build items are final (usually empty) classes which extend io.quarkus.builder.item.EmptyBuildItem.They represent build items that don’t actually carry any data, and allow such items to be produced and consumedwithout having to instantiate empty classes. They cannot themselves be instantiated.

Example of an empty build item

  1. public final class NativeImageBuildItem extends EmptyBuildItem {
  2. // empty
  3. }

Empty build items can represent "barriers" which can impose ordering between steps. They can also be used inthe same way that popular build systems use "pseudo-targets", which is to say that the build item can represent aconceptual goal that does not have a concrete representation.

Example of usage of an empty build item in a "pseudo-target" style

  1. /**
  2. * Contrived build step that produces the native image on disk. The main augmentation
  3. * step (which is run by Maven or Gradle) would be declared to consume this empty item,
  4. * causing this step to be run.
  5. */
  6. @BuildStep
  7. @Produces(NativeImageBuildItem.class)
  8. void produceNativeImage() {
  9. // ...
  10. // (produce the native image)
  11. // ...
  12. }

Example of usage of an empty build item in a "barrier" style

  1. /**
  2. * This would always run after {@link #produceNativeImage()} completes, producing
  3. * an instance of {@code SomeOtherBuildItem}.
  4. */
  5. @BuildStep
  6. @Consumes(NativeImageBuildItem.class)
  7. SomeOtherBuildItem secondBuildStep() {
  8. return new SomeOtherBuildItem("foobar");
  9. }

2.3.3. Injection

Classes which contain build steps support the following types of injection:

  • Constructor parameter injection

  • Field injection

  • Method parameter injection (for build step methods only)

Build step classes are instantiated and injected for each build step invocation, and are discarded afterwards. Stateshould only be communicated between build steps by way of build items, even if the steps are on the same class.

Final fields are not considered for injection, but can be populated by way of constructor parameter injectionif desired. Static fields are never considered for injection.

The types of values that can be injected include:

Objects which are injected into a build step method or its class must not be used outside of that method’sexecution.
Injection is resolved at compile time via an annotation processor,and the resulting code does not have permission to inject private fields or invoke private methods.

2.3.4. Producing values

A build step may produce values for subsequent steps in several possible ways:

  • By returning a simple build item or multi build item instance

  • By returning a List of a multi build item class

  • By injecting a BuildProducer of a simple or multi build item class

  • By annotating the method with @io.quarkus.deployment.annotations.Produces, giving the class name of aempty build item

If a simple build item is declared on a build step, it must be produced during that build step, otherwise an errorwill result. Build producers which are injected into steps must not be used outside of that step.

Note that a @BuildStep method will only be called if it produces something that another consumer or the final outputrequires. If there is no consumer for a particular item then it will not be produced. What is required will depend onthe final target that is being produced. For example, when running in developer mode the final output will not askfor GraalVM-specific build items such as ReflectiveClassBuildItem, so methods that only produce theseitems will not be invoked.

2.3.5. Consuming values

A build step may consume values from previous steps in the following ways:

  • By injecting a simple build item

  • By injecting an Optional of a simple build item class

  • By injecting a List of a multi build item class

  • By annotating the method with @io.quarkus.deployment.annotations.Consumes, giving the class name of aempty build item

Normally it is an error for a step which is included to consume a simple build item that is not produced by any otherstep. In this way, it is guaranteed that all of the declared values will be present and non-null when a step is run.

Sometimes a value isn’t necessary for the build to complete, but might inform some behavior of the build step if it ispresent. In this case, the value can be optionally injected.

Multi build values are always considered optional. If not present, an empty list will be injected.
2.3.5.1. Weak value production

Normally a build step is included whenever it produces any build item which is in turn consumed by any other build step. In this way,only the steps necessary to produce the final artifact(s) are included, and steps which pertain to extensions which arenot installed or which only produce build items which are not relevant for the given artifact type are excluded.

For cases where this is not desired behavior, the @io.quarkus.deployment.annotations.Weak annotation may be used. Thisannotation indicates that the build step should not automatically be included solely on the basis of producing the annotated value.

Example of producing a build item weakly

  1. /**
  2. * This build step is only run if something consumes the ExecutorClassBuildItem.
  3. */
  4. @BuildStep
  5. void createExecutor(
  6. @Weak BuildProducer<GeneratedClassBuildItem> classConsumer,
  7. BuildProducer<ExecutorClassBuildItem> executorClassConsumer
  8. ) {
  9. ClassWriter cw = new ClassWriter(Opcodes.ASM7);
  10. String className = generateClassThatCreatesExecutor(cw); (1)
  11. classConsumer.produce(new GeneratedClassBuildItem(true, className, cw.toByteArray()));
  12. executorClassConsumer.produce(new ExecutorClassBuildItem(className));
  13. }
1This method (not provided in this example) would generate the class using the ASM API.

Certain types of build items are generally always consumed, such as generated classes or resources.An extension might produce a build item along with a generated class to facilitate the usageof that build item. Such a build step would use the @Weak annotation on the generated class build item, while normallyproducing the other build item. If the other build item is ultimately consumed by something, then the step would runand the class would be generated. If nothing consumes the other build item, the step would not be included in the buildprocess.

In the example above, GeneratedClassBuildItem would only be produced if ExecutorClassBuildItem is consumed bysome other build step.

Note that when using bytecode recording, the implicitly generated class can be declared to be weak byusing the optional attribute of the @io.quarkus.deployment.annotations.Record annotation.

Example of using a bytecode recorder where the generated class is weakly produced

  1. /**
  2. * This build step is only run if something consumes the ExecutorBuildItem.
  3. */
  4. @BuildStep
  5. @Record(value = ExecutionTime.RUNTIME_INIT, optional = true) (1)
  6. ExecutorBuildItem createExecutor( (2)
  7. ExecutorTemplate executorTemplate,
  8. ThreadPoolConfig threadPoolConfig
  9. ) {
  10. return new ExecutorBuildItem(
  11. setupTemplate.setupRunTime(
  12. shutdownContextBuildItem,
  13. threadPoolConfig,
  14. launchModeBuildItem.getLaunchMode()
  15. )
  16. );
  17. }
1Note the optional attribute.
2This example is using recorder proxies; see the section on bytecode recording for more information.

2.3.6. Capabilities

The @BuildStep annotation has a providesCapabilities property that can be used to provide capability informationto other extensions about what is present in the current application. Capabilities are simply strings that are used todescribe an extension. Capabilities should generally be named after an extensions root package, for example the transactionsextension will provide io.quarkus.transactions.

To check if a capability is present you can inject the io.quarkus.deployment.Capabilities object and callisCapabilityPresent.

Capabilities should be used when checking for the presence of an extension rather than class path based checks.

2.3.7. Application Archives

The @BuildStep annotation can also register marker files that determine which archives on the class path are consideredto be 'Application Archives', and will therefore get indexed. This is done via the applicationArchiveMarkers. Forexample the ArC extension registers META-INF/beans.xml, which means that all archives on the class path with a beans.xmlfile will be indexed.

2.4. Configuration

Configuration in Quarkus is based on SmallRye Config, an implementation of the MicroProfile Config specification.All of the standard features of MP-Config are supported; in addition, there are several extensions which are made availableby the SmallRye Config project as well as by Quarkus itself.

The value of these properties is configured in a application.properties file that follows the MicroProfile config format.

Configuration of Quarkus extensions is injection-based, using annotations.

2.4.1. Configuration Keys

Leaf configuration keys are mapped to non-private fields via the @io.quarkus.runtime.annotations.ConfigItem annotation.

Though the SmallRye Config project is used for implementation, the standard @ConfigProperty annotation does not have thesame semantics that are needed to support configuration within extensions.

Configuration keys are normally derived from the field names that they are tied to. This is done by de-camel-casing the name and thenjoining the segments with hyphens (-). Some examples:

  • bindAddress becomes bind-address

  • keepAliveTime becomes keep-alive-time

  • requestDNSTimeout becomes request-dns-timeout

The name can also be explicitly specified by giving a name attribute to the @ConfigItem annotation.

Though it is possible to override the configuration key name using the name attribute of @ConfigItem,normally this should only be done in cases where (for example) the configuration key name is the same as a Java keyword.

2.4.2. Configuration Value types

The type of the field with the @ConfigItem annotation determines the conversion that is applied to it. Quarkusextensions may use the full range of configuration types made available by SmallRye Config, which includes:

  • All primitive types and primitive wrapper types

  • String

  • Any type which has a constructor accepting a single argument of type String or CharSequence

  • Any type which has a static method named of which accepts a single argument of type String

  • Any type which has a static method named valueOf or parse which accepts a single argument of type CharSequence or String

  • java.time.Duration

  • java.util.regex.Pattern

  • java.nio.file.Path

  • io.quarkus.runtime.configuration.MemorySize to represent data sizes

  • java.net.InetSocketAddress, java.net.InetAddress and org.wildfly.common.net.CidrAddress

  • A List or Optional of any of the above types

  • OptionalInt, OptionalLong, OptionalDouble

In addition, custom converters may be registered by adding their fully qualified class name in fileMETA-INF/services/org.eclipse.microprofile.config.spi.Converter.

Though these implicit converters use reflection, Quarkus will automatically ensure that they are loaded at the appropriate time.

2.4.3. Configuration Groups

Configuration values are always collected into grouping classes which are marked with the @io.quarkus.runtime.annotations.ConfigGroupannotation. These classes contain a field for each key within its group. In addition, configuration groups can be nested.

2.4.4. Configuration Maps

A Map can be used for configuration at any position where a configuration group would be allowed. The key type of such amap must be String, and its value may be either a configuration group class or a valid leaf type. The configurationkey segment following the map’s key segment will be used as the key for map values.

2.4.5. Configuration Roots

Configuration roots are configuration groups that appear in the root of the configuration tree. A configuration property’s fullname is determined by joining the string quarkus. with the hyphenated name of the fields that form the path from the root to theleaf field. For example, if I define a configuration root group called ThreadPool, with a nested group in a field named sizingthat in turn contains a field called minSize, the final configuration property will be called quarkus.thread-pool.sizing.min-size.

A configuration root’s name can be given with the name property, or it can be inferred from the class name. If the latter,then the configuration key will be the class name, minus any Config or Configuration suffix, broken up by camel-case,lowercased, and re-joined using hyphens (-).

A configuration root’s class name can contain an extra suffix segment for the case where there are configurationroots for multiple Configuration Root Phases. Classes which correspond to the BUILD_TIME and BUILD_AND_RUN_TIME_FIXEDmay end with BuildTimeConfig or BuildTimeConfiguration, and classes which correspond to the RUN_TIME phasemay end with RunTimeConfig or RunTimeConfiguration.

Note: The current implementation is still using injection site to determine the root set, so to avoid migration problems, itis recommended that the injection site (field or parameter) have the same name as the configuration root class untilthis change is complete.

2.4.5.1. Configuration Root Phases

A configuration root dictates when its contained keys are read from configuration, and when they are available to applications. The phases defined by io.quarkus.runtime.annotations.ConfigPhase are as follows:

Phase nameRead & avail. at build timeAvail. at run timeRead during static initRe-read during startup (native executable)Notes
BUILD_TIMEAppropriate for things which affect build.
BUILD_AND_RUN_TIME_FIXEDAppropriate for things which affect build and must be visible for run time code. Not read from config at run time.
RUN_TIMENot available at build, read at start in all modes.

For all cases other than the BUILD_TIME case, the configuration root class and all of the configuration groups and types contained therein must be located in, or reachable from, the extension’s run time artifact. Configuration roots of phase BUILD_TIME may be located in or reachable from either of the extension’s run time or deployment artifacts.

2.4.6. Configuration Example

  1. import io.quarkus.runtime.annotations.ConfigItem;
  2. import io.quarkus.runtime.annotations.ConfigGroup;
  3. import io.quarkus.runtime.annotations.DefaultConverter
  4. import java.io.File;
  5. import java.util.logging.Level;
  6. @ConfigGroup (1)
  7. public class FileConfig {
  8. /**
  9. * Enable logging to a file.
  10. */
  11. @ConfigItem(defaultValue = "true")
  12. boolean enable;
  13. /**
  14. * The log format.
  15. */
  16. @ConfigItem(defaultValue = "%d{yyyy-MM-dd HH:mm:ss,SSS} %h %N[%i] %-5p [%c{1.}] (%t) %s%e%n")
  17. String format;
  18. /**
  19. * The level of logs to be written into the file.
  20. */
  21. @DefaultConverter
  22. @ConfigItem(defaultValue = "ALL")
  23. Level level;
  24. /**
  25. * The name of the file in which logs will be written.
  26. */
  27. @ConfigItem(defaultValue = "application.log")
  28. File path;
  29. }
  30. /**
  31. * Logging configuration.
  32. */
  33. @ConfigRoot(phase = ConfigPhase.RUN_TIME) (2)
  34. public class LogConfiguration {
  35. // ...
  36. /**
  37. * Configuration properties for the logging file handler.
  38. */
  39. FileConfig file;
  40. }
  41. public class LoggingProcessor {
  42. // ...
  43. /**
  44. * Logging configuration.
  45. */
  46. (3)
  47. LogConfiguration config;
  48. }

A configuration property name can be split into segments. For example, a property name likequarkus.log.file.enable can be split into the following segments:

  • quarkus - a namespace claimed by Quarkus which is a prefix for all @ConfigRoot classes,

  • log - a name segment which corresponds to the LogConfiguration class annotated with @ConfigRoot,

  • file - a name segment which corresponds to the file field in this class,

  • enabled - a name segment which corresponds to enable field in FileConfig class annotated with @ConfigGroup.

1The FileConfig class is annotated with @ConfigGroup to indicate that this is an aggregateconfiguration object containing a collection of configurable properties, rather than being a simple configurationkey type.
2The @ConfigRoot annotation indicates that this object is a configuration root group, in this case one whichcorresponds to a log segment. A class name is used to link configuration root group with the segment from aproperty name. The Configuration part is stripped off from a LogConfiguration class name and the remaining Logis lowercased to become a log. Since all @ConfigRoot annotated classes uses quarkus as a prefix, this finallybecomes quarkus.log and represents the properties which names begin with quarkus.log..
*3Here the LoggingProcessor injects a LogConfiguration instance automatically by detecting the @ConfigRootannotation.

A corresponding application.properties for the above example could be:

  1. quarkus.log.file.enable=true
  2. quarkus.log.file.level=DEBUG
  3. quarkus.log.file.path=/tmp/debug.log

Since format is not defined in these properties, the default value from @ConfigItem will be used instead.

2.5. Conditional Step Inclusion

It is possible to only include a given @BuildStep under certain conditions. The @BuildStep annotationhas two optional parameters: onlyIf and onlyIfNot. These parameters can be set to one or more classeswhich implement BooleanSupplier. The build step will only be included when the method returnstrue (for onlyIf) or false (for onlyIfNot).

The condition class can inject configuration roots as long as they belong toa build-time phase. Run time configuration is not available for condition classes.

The condition class may also inject a value of type io.quarkus.runtime.LaunchMode.Constructor parameter and field injection is supported.

An example of a conditional build step

  1. @BuildStep(onlyIf = IsDevMode.class)
  2. LogCategoryBuildItem enableDebugLogging() {
  3. return new LogCategoryBuildItem("org.your.quarkus.extension", Level.DEBUG);
  4. }
  5. static class IsDevMode implements BooleanSupplier {
  6. LaunchMode launchMode;
  7. public boolean getAsBoolean() {
  8. return launchMode == LaunchMode.DEVELOPMENT;
  9. }
  10. }

2.5.1. Enhanced conversion

You can use enhanced conversion of a config item by using the @ConvertWith annotation which accepts a Converter class object.If the annotation is present on a config item, the implicit or custom built in converter in use will be overridden by the value provided.To do, see the example below which converts YES or NO values to boolean.

  1. @ConfigRoot
  2. public class SomeConfig {
  3. /**
  4. * Config item with enhanced converter
  5. */
  6. @ConvertWith(YesNoConverter.class) (1)
  7. @ConfigItem(defaultValue = "NO")
  8. Boolean answer;
  9. public static class YesNoConverter implements Converter<Boolean> {
  10. public YesNoConverter() {}
  11. @Override
  12. public Boolean convert(String s) {
  13. if (s == null || s.isEmpty()) {
  14. return false;
  15. }
  16. switch (s) {
  17. case "YES":
  18. return true;
  19. case "NO":
  20. return false;
  21. }
  22. throw new IllegalArgumentException("Unsupported value " + s + " given");
  23. }
  24. }
  25. }
  • Override the default Boolean converter and use the provided converter which accepts a YES or NO config values.

The corresponding application.properties will look like.

  1. quarkus.some.answer=YES
Enum values (config items) are translated to skewed-case (hyphenated) by default. The table below illustrates an enum name and their canonical equivalence:
Java enumCanonical equivalent
DISCARDdiscard
READ_UNCOMMITTEDread-uncommitted
SIGUSR1sigusr1
JavaEnumjava-enum
MAKING_LifeDifficultmaking-life-difficult
YeOldeJBossye-olde-jboss
camelCaseEnumcamel-case-enum

To use the default behaviour which is based on implicit converter or a custom defined one add @DefaultConverter annotation to the configuration item

  1. @ConfigRoot
  2. public class SomeLogConfig {
  3. /**
  4. * The level of logs to be written into the file.
  5. */
  6. @DefaultConverter (1)
  7. @ConfigItem(defaultValue = "ALL")
  8. Level level;
  9. }
  • Use the default converter (built in or a custom converter) to convert Level.class enum.

2.6. Bytecode Recording

One of the main outputs of the build process is recorded bytecode. This bytecode actually sets up the runtime environment. For example, in order to start Undertow, the resulting application will have some bytecode that directly registers allServlet instances and then starts Undertow.

As writing bytecode directly is complex, this is instead done via bytecode recorders. At deployment time,invocations are made on recorder objects that contain the actual runtime logic, but instead of these invocationsproceeding as normal they are intercepted and recorded (hence the name). This recording is then used to generate bytecodethat performs the same sequence of invocations at runtime. This is essentially a form of deferred execution where invocationsmade at deployment time get deferred until runtime.

Let’s look at the classic 'Hello World' type example. To do this the Quarkus way we would create a recorder as follows:

  1. @Recorder
  2. class HelloRecorder {
  3. public void sayHello(String name) {
  4. System.out.println("Hello" + name);
  5. }
  6. }

And then create a build step that uses this recorder:

  1. @Record(RUNTIME_INIT)
  2. @BuildStep
  3. public void helloBuildStep(HelloRecorder recorder) {
  4. recorder.sayHello("World");
  5. }

When this build step is run nothing is printed to the console. This is because the HelloRecorder that is injected isactually a proxy that records all invocations. Instead if we run the resulting Quarkus program we will see 'Hello World'printed to the console.

Methods on a recorder can return a value, which must be proxiable (if you want to return a non-proxiable item wrap itin io.quarkus.runtime.RuntimeValue). These proxies may not be invoked directly, however they can be passedinto other recorder methods. This can be any recorder method, including from other @BuildStep methods, so a common patternis to produce BuildItem instances that wrap the results of these recorder invocations.

For instance, in order to make arbitrary changes to a Servlet deployment Undertow has a ServletExtensionBuildItem,which is a MultiBuildItem that wraps a ServletExtension instance. I can return a ServletExtension from a recorderin another module, and Undertow will consume it and pass it into the recorder method that starts Undertow.

At runtime the bytecode will be invoked in the order it is generated. This means that build step dependencies implicitlycontrol the order that generated bytecode is run. In the example above we know that the bytecode that produces aServletExtensionBuildItem will be run before the bytecode that consumes it.

2.6.1. RecorderContext

io.quarkus.deployment.recording.RecorderContext provides some convenience methods to enhance bytecode recording,this includes the ability to register creation functions for classes without no-arg constructors, to register an objectsubstitution (basically a transformer from a non-serializable object to a serializable one and vice versa), and to createa class proxy. This interface can be directly injected as a method parameter into any @Record method.

Calling classProxy with a given class name will create a Class that can be passed into recordermethods, and at runtime will be substituted with the class whose name was passed in to classProxy. This is basically aconvenience to avoid the need to explicitly load classes in the recorders.

2.7. Contexts and Dependency Injection

2.7.1. Extension Points

As a CDI based runtime, Quarkus extensions often make CDI beans available as part of the extension behavior.However, Quarkus DI solution does not support CDI Portable Extensions.Instead, Quarkus extensions can make use of various Build Time Extension Points.

2.8. Extension Health Check

Health checks are provided via the quarkus-smallrye-health extension. It provides both liveness and readiness checks capabilities.

When writing an extension, it’s beneficial to provide health checks for the extension, that can be automatically included without the developer needing to write their own.

In order to provide a health check, you should do the following:

  • Import the quarkus-smallrye-health extension as an optional dependency in your runtime module so it will not impact the size of the application ifhealth check is not included.

  • Create your health check following the Quarkus - MicroProfile Health guide. We advise providing onlyreadiness check for an extension (liveness check is designed to express the fact that an application is up and needs to be lightweight).

  • Import the quarkus-smallrye-health-spi library in your deployment module.

  • Add a build step in your deployment module that produces a HealthBuildItem.

  • Add a way to disable the extension health check via a config item quarkus.<extension>.health.enabled that should be enabled by default.

Following is an example from the Agroal extension that provides a DataSourceHealthCheck to validate the readiness of a datasource.

  1. @BuildStep
  2. HealthBuildItem addHealthCheck(AgroalBuildTimeConfig agroalBuildTimeConfig) {
  3. return new HealthBuildItem("io.quarkus.agroal.runtime.health.DataSourceHealthCheck",
  4. agroalBuildTimeConfig.healthEnabled, "datasource");
  5. }

2.9. Customizing JSON handling from an extension

Extensions often need to register serializers and/or deserializers for types the extension provides.

For this, both JSON-B and Jackson extension provide a way to register serializer/deserializer from within anextension deployment module.

Keep in mind that not everybody will need JSON, so you need to make it optional.

If an extension intends to provide JSON related customization,it is strongly advised to provide customization for both JSON-B and Jackson.

2.9.1. Customizing JSON-B

First, add an optional dependency to quarkus-jsonb on your extension’s runtime module.

  1. <dependency>
  2. <groupId>io.quarkus</groupId>
  3. <artifactId>quarkus-jsonb</artifactId>
  4. <optional>true</optional>
  5. </dependency>

Then create a serializer and/or a deserializer for JSON-B, an example of which can be seen in the mongodb-panache extension.

  1. public class ObjectIdSerializer implements JsonbSerializer<ObjectId> {
  2. @Override
  3. public void serialize(ObjectId obj, JsonGenerator generator, SerializationContext ctx) {
  4. if (obj != null) {
  5. generator.write(obj.toString());
  6. }
  7. }
  8. }

Add a dependency to quarkus-jsonb-spi on your extension’s deployment module.

  1. <dependency>
  2. <groupId>io.quarkus</groupId>
  3. <artifactId>quarkus-jsonb-spi</artifactId>
  4. </dependency>

Add a build step to your processor to register the serializer via the JsonbSerializerBuildItem.

  1. @BuildStep
  2. JsonbSerializerBuildItem registerJsonbSerializer() {
  3. return new JsonbSerializerBuildItem(io.quarkus.mongodb.panache.jsonb.ObjectIdSerializer.class.getName()));
  4. }

The JSON-B extension will then use the produced build item to register your serializer/deserializer automatically.

If you need more customization capabilities than registering a serializer or a deserializer,you can produce a CDI bean that implements io.quarkus.jsonb.JsonbConfigCustomizer via an AdditionalBeanBuildItem.More info about customizing JSON-B can be found on the JSON guide Configuring JSON support

2.9.2. Customizing Jackson

First, add an optional dependency to quarkus-jackson on your extension’s runtime module.

  1. <dependency>
  2. <groupId>io.quarkus</groupId>
  3. <artifactId>quarkus-jackson</artifactId>
  4. <optional>true</optional>
  5. </dependency>

Then create a serializer or a deserializer (or both) for Jackson, an example of which can be seen in the mongodb-panache extension.

  1. public class ObjectIdSerializer extends StdSerializer<ObjectId> {
  2. public ObjectIdSerializer() {
  3. super(ObjectId.class);
  4. }
  5. @Override
  6. public void serialize(ObjectId objectId, JsonGenerator jsonGenerator, SerializerProvider serializerProvider)
  7. throws IOException {
  8. if (objectId != null) {
  9. jsonGenerator.writeString(objectId.toString());
  10. }
  11. }
  12. }

Add a dependency to quarkus-jackson-spi on your extension’s deployment module.

  1. <dependency>
  2. <groupId>io.quarkus</groupId>
  3. <artifactId>quarkus-jackson-spi</artifactId>
  4. </dependency>

Add a build step to your processor to register a Jackson module via the JacksonModuleBuildItem.You need to name your module in a unique way across all Jackson modules.

  1. @BuildStep
  2. JacksonModuleBuildItem registerJacksonSerDeser() {
  3. return new JacksonModuleBuildItem.Builder("ObjectIdModule")
  4. .add(io.quarkus.mongodb.panache.jackson.ObjectIdSerializer.class.getName(),
  5. io.quarkus.mongodb.panache.jackson.ObjectIdDeserializer.class.getName(),
  6. ObjectId.class.getName())
  7. .build();
  8. }

The Jackson extension will then use the produced build item to register a module within Jackson automatically.

If you need more customization capabilities than registering a module,you can produce a CDI bean that implements io.quarkus.jackson.ObjectMapperCustomizer via an AdditionalBeanBuildItem.More info about customizing Jackson can be found on the JSON guide Configuring JSON support

2.10. Testing Extensions

Testing of extensions should be done with the io.quarkus.test.QuarkusUnitTest runner. This runner allowsfor Arquillian-style tests that test specific functionalities. It is not intended for testing user applications, as thisshould be done via io.quarkus.test.junit.QuarkusTest. The main difference between these test runners is thatQuarkusTest simply boots the application once at the start of the run, while QuarkusUnitTest deploys a customQuarkus application for each test class.

These tests should be placed in the deployment module, if additional Quarkus modules are required for testingtheir deployment modules should also be added as test scoped dependencies.

Note that QuarkusUnitTest is in the quarkus-junit5-internal module.

An example test class may look like:

  1. package io.quarkus.health.test;
  2. import static org.junit.jupiter.api.Assertions.assertEquals;
  3. import java.util.ArrayList;
  4. import java.util.List;
  5. import javax.enterprise.inject.Instance;
  6. import javax.inject.Inject;
  7. import org.eclipse.microprofile.health.Health;
  8. import org.eclipse.microprofile.health.HealthCheck;
  9. import org.eclipse.microprofile.health.HealthCheckResponse;
  10. import io.quarkus.test.QuarkusUnitTest;
  11. import org.jboss.shrinkwrap.api.ShrinkWrap;
  12. import org.jboss.shrinkwrap.api.asset.EmptyAsset;
  13. import org.jboss.shrinkwrap.api.spec.JavaArchive;
  14. import org.junit.jupiter.api.Test;
  15. import org.junit.jupiter.api.extension.RegisterExtension;
  16. import io.restassured.RestAssured;
  17. public class FailingUnitTest {
  18. @RegisterExtension (1)
  19. static final QuarkusUnitTest config = new QuarkusUnitTest()
  20. .setArchiveProducer(() ->
  21. ShrinkWrap.create(JavaArchive.class) (2)
  22. .addClasses(FailingHealthCheck.class)
  23. .addAsManifestResource(EmptyAsset.INSTANCE, "beans.xml")
  24. );
  25. @Inject (3)
  26. @Health
  27. Instance<HealthCheck> checks;
  28. @Test
  29. public void testHealthServlet() {
  30. RestAssured.when().get("/health").then().statusCode(503); (4)
  31. }
  32. @Test
  33. public void testHealthBeans() {
  34. List<HealthCheck> check = new ArrayList<>(); (5)
  35. for (HealthCheck i : checks) {
  36. check.add(i);
  37. }
  38. assertEquals(1, check.size());
  39. assertEquals(HealthCheckResponse.State.DOWN, check.get(0).call().getState());
  40. }
  41. }
1This tells JUnit to use the Quarkus unit test runner
2This producer is used to build the application to be tested. It uses Shrinkwrap to create a JavaArchive to test
3It is possible to inject beans from our test deployment directly into the test case
4This method directly invokes the health check Servlet and verifies the response
5This method uses the injected health check bean to verify it is returning the expected result

If you want to test that an extension properly fails at build time, use the setExpectedException method:

  1. package io.quarkus.hibernate.orm;
  2. import io.quarkus.deployment.configuration.ConfigurationError;
  3. import io.quarkus.test.QuarkusUnitTest;
  4. import org.jboss.shrinkwrap.api.ShrinkWrap;
  5. import org.jboss.shrinkwrap.api.spec.JavaArchive;
  6. import org.junit.jupiter.api.Assertions;
  7. import org.junit.jupiter.api.Test;
  8. import org.junit.jupiter.api.extension.RegisterExtension;
  9. public class PersistenceAndQuarkusConfigTest {
  10. @RegisterExtension
  11. static QuarkusUnitTest runner = new QuarkusUnitTest()
  12. .setExpectedException(ConfigurationError.class) (1)
  13. .setArchiveProducer(() -> ShrinkWrap.create(JavaArchive.class)
  14. .addAsManifestResource("META-INF/some-persistence.xml", "persistence.xml")
  15. .addAsResource("application.properties"));
  16. @Test
  17. public void testPersistenceAndConfigTest() {
  18. // should not be called, deployment exception should happen first:
  19. // it's illegal to have Hibernate configuration properties in both the
  20. // application.properties and in the persistence.xml
  21. Assertions.fail();
  22. }
  23. }
1This tells JUnit that the Quarkus deployment should fail with a specific exception

2.11. Testing hot reload

It is also possible to write tests that verify an extension works correctly in development mode and can correctlyhandle updates.

For most extensions this will just work 'out of the box', however it is still a good idea to have a smoke test toverify that this functionality is working as expected. To test this we use QuarkusDevModeTest:

  1. public class ServletChangeTestCase {
  2. @RegisterExtension
  3. final static QuarkusDevModeTest test = new QuarkusDevModeTest()
  4. .setArchiveProducer(new Supplier<JavaArchive>() {
  5. @Override
  6. public JavaArchive get() {
  7. return ShrinkWrap.create(JavaArchive.class) (1)
  8. .addClass(DevServlet.class)
  9. .addAsManifestResource(new StringAsset("Hello Resource"), "resources/file.txt");
  10. }
  11. });
  12. @Test
  13. public void testServletChange() throws InterruptedException {
  14. RestAssured.when().get("/dev").then()
  15. .statusCode(200)
  16. .body(is("Hello World"));
  17. test.modifySourceFile("DevServlet.java", new Function<String, String>() { (2)
  18. @Override
  19. public String apply(String s) {
  20. return s.replace("Hello World", "Hello Quarkus");
  21. }
  22. });
  23. RestAssured.when().get("/dev").then()
  24. .statusCode(200)
  25. .body(is("Hello Quarkus"));
  26. }
  27. @Test
  28. public void testAddServlet() throws InterruptedException {
  29. RestAssured.when().get("/new").then()
  30. .statusCode(404);
  31. test.addSourceFile(NewServlet.class); (3)
  32. RestAssured.when().get("/new").then()
  33. .statusCode(200)
  34. .body(is("A new Servlet"));
  35. }
  36. @Test
  37. public void testResourceChange() throws InterruptedException {
  38. RestAssured.when().get("/file.txt").then()
  39. .statusCode(200)
  40. .body(is("Hello Resource"));
  41. test.modifyResourceFile("META-INF/resources/file.txt", new Function<String, String>() { (4)
  42. @Override
  43. public String apply(String s) {
  44. return "A new resource";
  45. }
  46. });
  47. RestAssured.when().get("file.txt").then()
  48. .statusCode(200)
  49. .body(is("A new resource"));
  50. }
  51. @Test
  52. public void testAddResource() throws InterruptedException {
  53. RestAssured.when().get("/new.txt").then()
  54. .statusCode(404);
  55. test.addResourceFile("META-INF/resources/new.txt", "New File"); (5)
  56. RestAssured.when().get("/new.txt").then()
  57. .statusCode(200)
  58. .body(is("New File"));
  59. }
  60. }
1This starts the deployment, your test can modify it as part of the test suite. Quarkus will be restarted betweeneach test method so every method starts with a clean deployment.
2This method allows you to modify the source of a class file. The old source is passed into the function, and the updatedsource is returned.
3This method adds a new class file to the deployment. The source that is used will be the original source that is partof the current project.
4This method modifies a static resource
5This method adds a new static resource

2.12. Native Executable Support

There Quarkus provides a lot of build items that control aspects of the native executable build. This allows for extensionsto programmatically perform tasks such as registering classes for reflection or adding static resources to the nativeexecutable. Some of these build items are listed below:

  • io.quarkus.deployment.builditem.nativeimage.NativeImageResourceBuildItem
  • Includes static resources into the native executable.

  • io.quarkus.deployment.builditem.nativeimage.RuntimeReinitializedClassBuildItem

  • A class that will be reinitialized at runtime by Substrate. This will result in the static initializer running twice.

  • io.quarkus.deployment.builditem.nativeimage.NativeImageSystemPropertyBuildItem

  • A system property that will be set at native executable build time.

  • io.quarkus.deployment.builditem.nativeimage.NativeImageResourceBundleBuildItem

  • Includes a resource bundle in the native executable.

  • io.quarkus.deployment.builditem.nativeimage.ReflectiveClassBuildItem

  • Registers a class for reflection in Substrate. Constructors are always registered, while methods and fields are optional.

  • io.quarkus.deployment.builditem.nativeimage.RuntimeInitializedClassBuildItem

  • A class that will be initialized at runtime rather than build time. This will cause the build to fail if the class is initialized as part of the native executable build process, so care must be taken.

  • io.quarkus.deployment.builditem.nativeimage.NativeImageConfigBuildItem

  • A convenience feature that allows you to control most of the above features from a single build item.

  • io.quarkus.deployment.builditem.NativeEnableAllCharsetsBuildItem

  • Indicates that all charsets should be enabled in native image.

  • io.quarkus.deployment.builditem.ExtensionSslNativeSupportBuildItem

  • A convenient way to tell Quarkus that the extension requires SSL and it should be enabled during native image build.When using this feature, remember to add your extension to the list of extensions that offer SSL support automatically on the native and ssl guide.

2.13. IDE support tips

2.13.1. Writing Quarkus extensions in Eclipse

The only particular aspect of writing Quarkus extensions in Eclipse is that APT (Annotation Processing Tool) is required as part of extension builds, which means you need to:

  • Install m2e-apt from https://marketplace.eclipse.org/content/m2e-apt

  • Define this property in your pom.xml: <m2e.apt.activation>jdt_apt</m2e.apt.activation>, although if you rely on io.quarkus:quarkus-build-parent you will get it for free.

  • If you have the io.quarkus:quarkus-extension-processor project open at the same time in your IDE (for example, if you have the Quarkus sources checked out and open in your IDE) you will need to close that project. Otherwise, Eclipse will not invoke the APT plugin that it contains.

  • If you just closed the extension processor project, be sure to do Maven > Update Project on the other projects in order for Eclipse to pick up the extension processor from the Maven repository.

2.14. Troubleshooting / Debugging Tips

2.14.1. Saving Application Generated Classes to Disk

The class augmentation step of Quarkus generates classes for various purposes. Sometimes you need to view theseclasses/bytecode to debug or understand an issue. Classes that are related to application augmentation are currently held inmemory in a runtime class loader.

When using the dev mode (running mvn clean compile quarkus:dev) you can have these classes written out to disk for inspection, for this, you need to specify thequarkus.debug.generated-classes-dir system property, for example:

./mvnw clean compile quarkus:dev -Dquarkus.debug.generated-classes-dir=app-generated-classes

You will see a line of log for each class outputed to the directory:

2019-10-25 18:12:34,262 INFO  [io.qua.run.RuntimeClassLoader] (main) Wrote /path/to/my/app/target/generated-devmode/io.quarkus.runner.GeneratedMain.class

2.14.2. Multi-module Maven Projects and the Development Mode

It’s not uncommon to develop an extension in a multi-module Maven project that also contains an "example" module.However, if you want to run the example in the development mode then the -DnoDeps system property must be used in order to exclude the local project dependencies.Otherwise, Quarkus attempts to monitor the extension classes and this may result in weird class loading issues.

./mvnw compile quarkus:dev -DnoDeps

2.15. Sample Test Extension

We have an extension that is used to test for regressions in the extension processing. It is located in https://github.com/quarkusio/quarkus/tree/master/core/test-extension directory. In this section we touch on some of the tasks an extensionauthor will typically need to perform using the test-extension code to illustrate how the task could be done.

2.15.1. Features and Capabilities

When you start a Quarkus instance, you will see a line like the one highlighted in this example:

Example Startup Lines

2019-03-22 14:02:37,884 INFO  [io.quarkus] (main) Quarkus 999-SNAPSHOT started in 0.061s.
2019-03-22 14:02:37,884 INFO  [io.quarkus] (main) Installed features: [cdi, test-extension] (1)
1A list of features installed in the runtime image

The features listed reflect the types of extensions that are installed. An extension declares its display name using a Build Step Processors method that produces a FeatureBuildItem like this TestProcessor#featureBuildItem() method example:

TestProcessor#featureBuildItem()

    /**
     * Register a extension capability and feature
     *
     * @return test-extension feature build item
     */
    @BuildStep(providesCapabilities = "io.quarkus.test-extension")
    FeatureBuildItem featureBuildItem() {
        return new FeatureBuildItem("test-extension");
    }

The feature name should map to a label in the extension’s devtools/common/src/main/filtered/extensions.json entry so thatthe feature name displayed by the startup line matches a label that one can used to select the extension when creating a projectusing the Quarkus maven plugin as shown in this example taken from the Writing JSON REST Services guide where the "resteasy-jsonb" feature is referenced:

mvn io.quarkus:quarkus-maven-plugin:1.0.0.CR1:create \
    -DprojectGroupId=org.acme \
    -DprojectArtifactId=rest-json \
    -DclassName="org.acme.rest.json.FruitResource" \
    -Dpath="/fruits" \
    -Dextensions="resteasy-jsonb"
cd rest-json

2.15.2. Bean Defining Annotations

The CDI layer processes CDI beans that are either explicitly registered or that it discovers based on bean defining annotations as defined in 2.5.1. Bean defining annotations. You can expand this set of annotations to include annotations your extension processes using a BeanDefiningAnnotationBuildItem as shown in this TestProcessor#registerBeanDefinningAnnotations example:

Register a Bean Definining Annotation

import javax.enterprise.context.ApplicationScoped;
import org.jboss.jandex.DotName;
import io.quarkus.extest.runtime.TestAnnotation;

public final class TestProcessor {
    static DotName TEST_ANNOTATION = DotName.createSimple(TestAnnotation.class.getName());
    static DotName TEST_ANNOTATION_SCOPE = DotName.createSimple(ApplicationScoped.class.getName());

...

    @BuildStep
    BeanDefiningAnnotationBuildItem registerX() {
        (1)
        return new BeanDefiningAnnotationBuildItem(TEST_ANNOTATION, TEST_ANNOTATION_SCOPE);
    }
...
}

/**
 * Marker annotation for test configuration target beans
 */
@Target({ TYPE })
@Retention(RUNTIME)
@Documented
@Inherited
public @interface TestAnnotation {
}

/**
 * A sample bean
 */
@TestAnnotation (2)
public class ConfiguredBean implements IConfigConsumer {

...
1Register the annotation class and CDI default scope using the Jandex DotName class.
2ConfiguredBean will be processed by the CDI layer the same as a bean annotated with the CDI standard @ApplicationScoped.

2.15.3. Parsing Config to Objects

One of the main things an extension is likely to do is completely separate the configuration phase of behavior from the runtime phase. Frameworks often do parsing/load of configuration on startup that can be done during build time to both reduce the runtime dependencies on frameworks like xml parsers as well as reducing the startup time the parsing incurs.

An example of parsing a XML config file using JAXB is shown in the TestProcessor#parseServiceXmlConfig method:.Parsing an XML Configuration into Runtime XmlConfig Instance

    @BuildStep
    @Record(STATIC_INIT)
    RuntimeServiceBuildItem parseServiceXmlConfig(TestRecorder recorder) throws JAXBException {
        RuntimeServiceBuildItem serviceBuildItem = null;
        JAXBContext context = JAXBContext.newInstance(XmlConfig.class);
        Unmarshaller unmarshaller = context.createUnmarshaller();
        InputStream is = getClass().getResourceAsStream("/config.xml"); (1)
        if (is != null) {
            log.infof("Have XmlConfig, loading");
            XmlConfig config = (XmlConfig) unmarshaller.unmarshal(is); (2)
...
        }
        return serviceBuildItem;
    }
1Look for a config.xml classpath resource
2If found, parse using JAXB context for XmlConfig.class
If there was no /config.xml resource available in the build environment, then a null RuntimeServiceBuildItem would be returned and no subsequent logic based on a RuntimeServiceBuildItem being produced would execute.

Typically one is loading a configuration to create some runtime component/service as parseServiceXmlConfig is doing. We will come back to the rest of the behavior in parseServiceXmlConfig in the following Manage Non-CDI Service section.

If for some reason you need to parse the config and use it in other build steps in an extension processor, you would need to create an XmlConfigBuildItem to pass the parsed XmlConfig instance around.

If you look at the XmlConfig code you will see that it does carry around the JAXB annotations. If you don’t want these in the runtime image, you could clone the XmlConfig instance into some POJO object graph and then replace XmlConfig with the POJO class. We will do this in Replacing Classes in the Native Image.

2.15.4. Scanning Deployments Using Jandex

If your extension defines annotations or interfaces that mark beans needing to be processed, you can locate these beans using the Jandex API, a Java annotation indexer and offline reflection library. The following TestProcessor#scanForBeans method shows how to find the beans annotated with our @TestAnnotation that also implement the IConfigConsumer interface:

Example Jandex Usage

    static DotName TEST_ANNOTATION = DotName.createSimple(TestAnnotation.class.getName());
...

    @BuildStep
    @Record(STATIC_INIT)
    void scanForBeans(TestRecorder recorder, BeanArchiveIndexBuildItem beanArchiveIndex, (1)
            BuildProducer<TestBeanBuildItem> testBeanProducer) {
        IndexView indexView = beanArchiveIndex.getIndex(); (2)
        Collection<AnnotationInstance> testBeans = indexView.getAnnotations(TEST_ANNOTATION); (3)
        for (AnnotationInstance ann : testBeans) {
            ClassInfo beanClassInfo = ann.target().asClass();
            try {
                boolean isConfigConsumer = beanClassInfo.interfaceNames()
                        .stream()
                        .anyMatch(dotName -> dotName.equals(DotName.createSimple(IConfigConsumer.class.getName()))); (4)
                if (isConfigConsumer) {
                    Class<IConfigConsumer> beanClass = (Class<IConfigConsumer>) Class.forName(beanClassInfo.name().toString());
                    testBeanProducer.produce(new TestBeanBuildItem(beanClass)); (5)
                    log.infof("Configured bean: %s", beanClass);
                }
            } catch (ClassNotFoundException e) {
                log.warn("Failed to load bean class", e);
            }
        }
    }
1Depend on a BeanArchiveIndexBuildItem to have the build step be run after the deployment has been indexed.
2Retrieve the index.
3Find all beans annotated with @TestAnnotation.
4Determine which of these beans also has the IConfigConsumer interface.
5Save the bean class in a TestBeanBuildItem for use in a latter RUNTIME_INIT build step that will interact with the bean instances.

2.15.5. Interacting With Extension Beans

You can use the io.quarkus.arc.runtime.BeanContainer interface to interact with your extension beans. The following configureBeans methods illustrate interacting with the beans scanned for in the previous section:

Using CDI BeanContainer Interface

// TestProcessor#configureBeans
    @BuildStep
    @Record(RUNTIME_INIT)
    void configureBeans(TestRecorder recorder, List<TestBeanBuildItem> testBeans, (1)
            BeanContainerBuildItem beanContainer, (2)
            TestRunTimeConfig runTimeConfig) {

        for (TestBeanBuildItem testBeanBuildItem : testBeans) {
            Class<IConfigConsumer> beanClass = testBeanBuildItem.getConfigConsumer();
            recorder.configureBeans(beanContainer.getValue(), beanClass, buildAndRunTimeConfig, runTimeConfig); (3)
        }
    }

// TestRecorder#configureBeans
    public void configureBeans(BeanContainer beanContainer, Class<IConfigConsumer> beanClass,
            TestBuildAndRunTimeConfig buildTimeConfig,
            TestRunTimeConfig runTimeConfig) {
        log.infof("Begin BeanContainerListener callback\n");
        IConfigConsumer instance = beanContainer.instance(beanClass); (4)
        instance.loadConfig(buildTimeConfig, runTimeConfig); (5)
        log.infof("configureBeans, instance=%s\n", instance);
    }
1Consume the TestBeanBuildItems produced from the scanning build step.
2Consume the BeanContainerBuildItem to order this build step to run after the CDI bean container has been created.
3Call the runtime recorder to record the bean interactions.
4Runtime recorder retrieves the bean using its type.
5Runtime recorder invokes the IConfigConsumer#loadConfig(…​) method passing in the configuration objects with runtime information.

2.15.6. Manage Non-CDI Service

A common purpose for an extension is to integrate a non-CDI aware service into the CDI based Quarkus runtime. Step 1 of this task is to load any configuration needed in a STATIC_INIT build step as we did in Parsing Config to Objects. Now we need to create an instance of the service using the configuration. Let’s return to the TestProcessor#parseServiceXmlConfig method to see how this can be done.

Creating a Non-CDI Service

// TestProcessor#parseServiceXmlConfig
    @BuildStep
    @Record(STATIC_INIT)
    RuntimeServiceBuildItem parseServiceXmlConfig(TestRecorder recorder) throws JAXBException {
        RuntimeServiceBuildItem serviceBuildItem = null;
        JAXBContext context = JAXBContext.newInstance(XmlConfig.class);
        Unmarshaller unmarshaller = context.createUnmarshaller();
        InputStream is = getClass().getResourceAsStream("/config.xml");
        if (is != null) {
            log.infof("Have XmlConfig, loading");
            XmlConfig config = (XmlConfig) unmarshaller.unmarshal(is);
            log.infof("Loaded XmlConfig, creating service");
            RuntimeValue<RuntimeXmlConfigService> service = recorder.initRuntimeService(config); (1)
            serviceBuildItem = new RuntimeServiceBuildItem(service); (3)
        }
        return serviceBuildItem;
    }

// TestRecorder#initRuntimeService
    public RuntimeValue<RuntimeXmlConfigService> initRuntimeService(XmlConfig config) {
        RuntimeXmlConfigService service = new RuntimeXmlConfigService(config); (2)
        return new RuntimeValue<>(service);
    }

// RuntimeServiceBuildItem
    final public class RuntimeServiceBuildItem extends SimpleBuildItem {
    private RuntimeValue<RuntimeXmlConfigService> service;

    public RuntimeServiceBuildItem(RuntimeValue<RuntimeXmlConfigService> service) {
        this.service = service;
    }

    public RuntimeValue<RuntimeXmlConfigService> getService() {
        return service;
    }
}
1Call into the runtime recorder to record the creation of the service.
2Using the parsed XmlConfig instance, create an instance of RuntimeXmlConfigService and wrap it in a RuntimeValue. Use a RuntimeValue wrapper for non-interface objects that are non-proxiable.
3Wrap the return service value in a RuntimeServiceBuildItem for use in a RUNTIME_INIT build step that will start the service.
2.15.6.1. Starting a Service

Now that you have recorded the creation of a service during the build phase, you need to record how to start the service at runtime during booting. You do this with a RUNTIME_INIT build step as shown in the TestProcessor#startRuntimeService method.

Starting/Stopping a Non-CDI Service

// TestProcessor#startRuntimeService
    @BuildStep
    @Record(RUNTIME_INIT)
    ServiceStartBuildItem startRuntimeService(TestRecorder recorder, ShutdownContextBuildItem shutdownContextBuildItem , (1)
            RuntimeServiceBuildItem serviceBuildItem) throws IOException { (2)
        if (serviceBuildItem != null) {
            log.info("Registering service start");
            recorder.startRuntimeService(shutdownContextBuildItem, serviceBuildItem.getService()); (3)
        } else {
            log.info("No RuntimeServiceBuildItem seen, check config.xml");
        }
        return new ServiceStartBuildItem("RuntimeXmlConfigService"); (4)
    }

// TestRecorder#startRuntimeService
    public void startRuntimeService(ShutdownContext shutdownContext, RuntimeValue<RuntimeXmlConfigService> runtimeValue)
            throws IOException {
        RuntimeXmlConfigService service = runtimeValue.getValue();
        service.startService(); (5)
        shutdownContext.addShutdownTask(service::stopService); (6)
    }
1We consume a ShutdownContextBuildItem to register the service shutdown.
2We consume the previously initialized service captured in RuntimeServiceBuildItem.
3Call the runtime recorder to record the service start invocation.
4Produce a ServiceStartBuildItem to indicate the startup of a service. See Startup and Shutdown Events for details.
5Runtime recorder retrieves the service instance reference and calls its startService method.
6Runtime recorder registers an invocation of the service instance stopService method with the Quarkus ShutdownContext.

The code for the RuntimeXmlConfigService can be viewed here:{quarkus-blob-url}/core/test-extension/runtime/src/main/java/io/quarkus/extest/runtime/RuntimeXmlConfigService.java[RuntimeXmlConfigService.java]

The testcase for validating that the RuntimeXmlConfigService has started can be found in the testRuntimeXmlConfigService test of ConfiguredBeanTest and NativeImageIT.

2.15.7. Startup and Shutdown Events

The Quarkus container supports startup and shutdown lifecycle events to notify components of the container startupand shutdown. There are CDI events fired that components can observe are illustrated in this example:

Observing Container Startup

import io.quarkus.runtime.ShutdownEvent;
import io.quarkus.runtime.StartupEvent;

puclic class SomeBean {
    /**
     * Called when the runtime has started
     * @param event
     */
    void onStart(@Observes StartupEvent event) { (1)
        System.out.printf("onStart, event=%s%n", event);
    }

    /**
     * Called when the runtime is shutting down
     * @param event
    */
    void onStop(@Observes ShutdownEvent event) { (2)
        System.out.printf("onStop, event=%s%n", event);
    }
}
1Observe a StartupEvent to be notified the runtime has started.
2Observe a 'ShutdownEvent` to be notified when the runtime is going to shutdown.

What is the relevance of startup and shutdown events for extension authors? We have already seen the use of a ShutdownContextto register a callback to perform shutdown tasks in the Starting a Service section. These shutdown tasks would be calledafter a ShutdownEvent had been sent.

A StartupEvent is fired after all io.quarkus.deployment.builditem.ServiceStartBuildItem producers have been consumed.The implication of this is that if an extension has services that application components would expect to have beenstarted when they observe a StartupEvent, the build steps that invoke the runtime code to start those services needsto produce a ServiceStartBuildItem to ensure that the runtime code is run before the StartupEvent is sent. Recall thatwe saw the production of a ServiceStartBuildItem in the previous section, and it is repeated here for clarity:

Example of Producing a ServiceStartBuildItem

// TestProcessor#startRuntimeService
    @BuildStep
    @Record(RUNTIME_INIT)
    ServiceStartBuildItem startRuntimeService(TestRecorder recorder, ShutdownContextBuildItem shutdownContextBuildItem,
            RuntimeServiceBuildItem serviceBuildItem) throws IOException {
...
        return new ServiceStartBuildItem("RuntimeXmlConfigService"); (1)
    }
1Produce a ServiceStartBuildItem to indicate that this is a service starting step that needs to run before the StartupEvent is sent.

2.15.8. Register Resources for Use in Native Image

Not all configuration or resources can be consumed at build time. If you have classpath resources that the runtime needs to access, you need to inform the build phase that these resources need to be copied into the native image. This is done by producing one or more NativeImageResourceBuildItem or NativeImageResourceBundleBuildItem in the case of resource bundles. Examples of this are shown in this sample registerNativeImageReources build step:

Registering Resources and ResourceBundles

public final class MyExtProcessor {
    @Inject
    BuildProducer<NativeImageResourceBuildItem> resource;
    @Inject
    BuildProducer<NativeImageResourceBundleBuildItem> resourceBundle;

    @BuildStep
    void registerNativeImageReources() {
        resource.produce(new NativeImageResourceBuildItem("/security/runtime.keys")); (1)

        resource.produce(new NativeImageResourceBuildItem(
                "META-INF/services/" + io.quarkus.SomeService.class.getName())); (2)

        resourceBundle.produce(new NativeImageResourceBuildItem("javax.xml.bind.Messages")); (3)
    }
}
1Indicate that the /security/runtime.keys classpath resource should be copied into native image.
2Indicate that the io.quarkus.SomeService ServiceLoader resource should be copied into native image.
3Indicate that the "javax.xml.bind.Messages" resource bundle should be copied into native image.

2.15.9. Object Substitution

Objects created during the build phase that are passed into the runtime need to have a default constructor in order for them to be created and configured at startup of the runtime from the build time state. If an object does not have a default constructor you will see an error similar to the following during generation of the augmented artifacts:

DSAPublicKey Serialization Error

    [error]: Build step io.quarkus.deployment.steps.MainClassBuildStep#build threw an exception: java.lang.RuntimeException: Unable to serialize objects of type class sun.security.provider.DSAPublicKeyImpl to bytecode as it has no default constructor
    at io.quarkus.builder.Execution.run(Execution.java:123)
    at io.quarkus.builder.BuildExecutionBuilder.execute(BuildExecutionBuilder.java:136)
    at io.quarkus.deployment.QuarkusAugmentor.run(QuarkusAugmentor.java:110)
    at io.quarkus.runner.RuntimeRunner.run(RuntimeRunner.java:99)
    ... 36 more

There is a io.quarkus.runtime.ObjectSubstitution interface that can be implemented to tell Quarkus how to handle such classes. An example implementation for the DSAPublicKey is shown here:

DSAPublicKeyObjectSubstitution Example

package io.quarkus.extest.runtime.subst;

import java.security.KeyFactory;
import java.security.NoSuchAlgorithmException;
import java.security.interfaces.DSAPublicKey;
import java.security.spec.InvalidKeySpecException;
import java.security.spec.X509EncodedKeySpec;
import java.util.logging.Logger;

import io.quarkus.runtime.ObjectSubstitution;

public class DSAPublicKeyObjectSubstitution implements ObjectSubstitution<DSAPublicKey, KeyProxy> {
    private static final Logger log = Logger.getLogger("DSAPublicKeyObjectSubstitution");
    @Override
    public KeyProxy serialize(DSAPublicKey obj) { (1)
        log.info("DSAPublicKeyObjectSubstitution.serialize");
        byte[] encoded = obj.getEncoded();
        KeyProxy proxy = new KeyProxy();
        proxy.setContent(encoded);
        return proxy;
    }

    @Override
    public DSAPublicKey deserialize(KeyProxy obj) { (2)
        log.info("DSAPublicKeyObjectSubstitution.deserialize");
        byte[] encoded = obj.getContent();
        X509EncodedKeySpec publicKeySpec = new X509EncodedKeySpec(encoded);
        DSAPublicKey dsaPublicKey = null;
        try {
            KeyFactory kf = KeyFactory.getInstance("DSA");
            dsaPublicKey = (DSAPublicKey) kf.generatePublic(publicKeySpec);

        } catch (NoSuchAlgorithmException | InvalidKeySpecException e) {
            e.printStackTrace();
        }
        return dsaPublicKey;
    }
}
1The serialize method takes the object without a default constructor and creates a KeyProxy that contains the information necessary to recreate the DSAPublicKey.
2The deserialize method uses the KeyProxy to recreate the DSAPublicKey from its encoded form using the key factory.

An extension registers this substitution by producing an ObjectSubstitutionBuildItem as shown in this TestProcessor#loadDSAPublicKey fragment:

Registering an Object Subtitution

    @BuildStep
    @Record(STATIC_INIT)
    PublicKeyBuildItem loadDSAPublicKey(TestRecorder recorder,
            BuildProducer<ObjectSubstitutionBuildItem> substitutions) throws IOException, GeneralSecurityException {
...
        // Register how to serialize DSAPublicKey
        ObjectSubstitutionBuildItem.Holder<DSAPublicKey, KeyProxy> holder = new ObjectSubstitutionBuildItem.Holder(
                DSAPublicKey.class, KeyProxy.class, DSAPublicKeyObjectSubstitution.class);
        ObjectSubstitutionBuildItem keysub = new ObjectSubstitutionBuildItem(holder);
        substitutions.produce(keysub);

        log.infof("loadDSAPublicKey run");
        return new PublicKeyBuildItem(publicKey);
    }

2.15.10. Replacing Classes in the Native Image

The Graal SDK supports substitutions of classes in the native image. An example of how one could replace the XmlConfig/XmlData classes with versions that have no JAXB annotation dependencies is shown in these example classes:

Substitution of XmlConfig/XmlData Classes Example

package io.quarkus.extest.runtime.graal;
import java.util.Date;
import com.oracle.svm.core.annotate.Substitute;
import com.oracle.svm.core.annotate.TargetClass;
import io.quarkus.extest.runtime.config.XmlData;

@TargetClass(XmlConfig.class)
@Substitute
public final class Target_XmlConfig {

    @Substitute
    private String address;
    @Substitute
    private int port;
    @Substitute
    private ArrayList<XData> dataList;

    @Substitute
    public String getAddress() {
        return address;
    }

    @Substitute
    public int getPort() {
        return port;
    }

    @Substitute
    public ArrayList<XData> getDataList() {
        return dataList;
    }

    @Substitute
    @Override
    public String toString() {
        return "Target_XmlConfig{" +
                "address='" + address + '\'' +
                ", port=" + port +
                ", dataList=" + dataList +
                '}';
    }
}

@TargetClass(XmlData.class)
@Substitute
public final class Target_XmlData {

    @Substitute
    private String name;
    @Substitute
    private String model;
    @Substitute
    private Date date;

    @Substitute
    public String getName() {
        return name;
    }

    @Substitute
    public String getModel() {
        return model;
    }

    @Substitute
    public Date getDate() {
        return date;
    }

    @Substitute
    @Override
    public String toString() {
        return "Target_XmlData{" +
                "name='" + name + '\'' +
                ", model='" + model + '\'' +
                ", date='" + date + '\'' +
                '}';
    }
}

3. Configuration reference documentation

The configuration is an important part of each extension and therefore needs to be properly documented.Each configuration property should have a proper Javadoc comment.

While it is handy to have the documentation available when coding, this configuration documentation must also be available in the extension guides.The Quarkus build automatically generates the configuration documentation for you based on the Javadoc comments but you need to explicitly include it in your guide.

In this section, we will explain everything you need to know about the configuration reference documentation.

3.1. Writing the documentation

For each configuration property, you need to write some Javadoc explaining its purpose.

Always make the first sentence meaningful and self-contained as it is included in the summary table.

You can either use standard Javadoc comments or Asciidoc directly as a Javadoc comment.

We assume you are familiar with writing Javadoc comments so let’s focus on our Asciidoc support.While standard Javadoc comments are perfectly fine for simple documentation (recommended even),if you want to include tips, source code extracts, lists…​ Asciidoc comes in handy.

Here is a typical configuration property commented with Asciidoc:

/**
 * Class name of the Hibernate ORM dialect. The complete list of bundled dialects is available in the
 * https://docs.jboss.org/hibernate/stable/orm/javadocs/org/hibernate/dialect/package-summary.html[Hibernate ORM JavaDoc].
 *
 * [NOTE]
 * ====
 * Not all the dialects are supported in GraalVM native executables: we currently provide driver extensions for PostgreSQL,
 * MariaDB, Microsoft SQL Server and H2.
 * ====
 *
 * @asciidoclet
 */
@ConfigItem
public Optional<String> dialect;

This is the simple case: you just have to write Asciidoc and mark the comment with the @asciidoclet tag.This tag has two purposes: it is used as a marker for our generation tool but it is also used by the javadoc process for proper Javadoc generation.

Now let’s consider a more complicated example:

// @formatter:off
/**
 * Name of the file containing the SQL statements to execute when Hibernate ORM starts.
 * Its default value differs depending on the Quarkus launch mode:
 *
 * * In dev and test modes, it defaults to `import.sql`.
 *   Simply add an `import.sql` file in the root of your resources directory
 *   and it will be picked up without having to set this property.
 *   Pass `no-file` to force Hibernate ORM to ignore the SQL import file.
 * * In production mode, it defaults to `no-file`.
 *   It means Hibernate ORM won't try to execute any SQL import file by default.
 *   Pass an explicit value to force Hibernate ORM to execute the SQL import file.
 *
 * If you need different SQL statements between dev mode, test (`@QuarkusTest`) and in production, use Quarkus
 * https://quarkus.io/guides/config#configuration-profiles[configuration profiles facility].
 *
 * [source,property]
 * .application.properties
 * ----
 * %dev.quarkus.hibernate-orm.sql-load-script = import-dev.sql
 * %test.quarkus.hibernate-orm.sql-load-script = import-test.sql
 * %prod.quarkus.hibernate-orm.sql-load-script = no-file
 * ----
 *
 * [NOTE]
 * ====
 * Quarkus supports `.sql` file with SQL statements or comments spread over multiple lines.
 * Each SQL statement must be terminated by a semicolon.
 * ====
 *
 * @asciidoclet
 */
// @formatter:on
@ConfigItem
public Optional<String> sqlLoadScript;

A few comments on this one:

  • Every time you will need the indentation to be respected in the Javadoc comment (think list items spread on multiple lines or indented source code),you will need to disable temporarily the automatic Eclipse formatter(this, even if you don’t use Eclipse as the formatter is included in our build).To do so, use the // @formatter:off/// @formatter:on markers.Note the fact that they are separate comments and there is a space after the // marker. This is required.

  • As you can see, you can use the full power of Asciidoctor (except for the limitation below)

You cannot use open blocks () in your Asciidoctor documentation.All the other types of blocks (source, admonitions…​) are supported.
By default, the doc generator will use the hyphenated field name as the key of a java.util.Map configuration item.To override this default and have a user friendly key (independent of implementation details), you may use the io.quarkus.runtime.annotations.ConfigDocMapKey annotation.See the following example,
@ConfigRootpublic class SomeConfig {    /*
      Namespace configuration.     */    @ConfigItem(name = ConfigItem.PARENT)    @ConfigDocMapKey("cache-name") (1)    Map<String, CaffeineNamespaceConfig> namespace;}
-This will generate a configuration map key named quarkus.some."cache-name" instead of quarkus.some."namespace".

3.2. Writing section documentation

If you wish to generate configuration section of a given @ConfigGroup, Quarkus has got you covered with the @ConfigDocSection annotation.See the code example below:

/**
* Config group related configuration.
* Amazing introduction here
*/
@ConfigItem
@ConfigDocSection (1)
public ConfigGroupConfig configGroup;
  • This will add a section documentation for the confgiGroup config item in the generated documentation.Section’s title and introduction will be derived from the javadoc of the configuration item. The first sentence from the javadoc is considered as the section title and the remaining sentences used as section introduction.You can also use the @asciidoclet tag as shown above.

3.3. Generating the documentation

Generating the documentation is easy:

  • Running ./mvnw clean install -DskipTests -DskipITs will do.

  • You can either do it globally or in a specific extension directory (e.g. extensions/mailer).

The documentation is generated in the global target/asciidoc/generated/config/ located at the root of the project.

3.4. Including the documentation in the extension guide

Now that you have generated the configuration reference documentation for your extension, you need to include it in your guide (and review it).

This is simple, include the generated documentation in your guide:

include::{generated-dir}/config/quarkus-your-extension.adoc[opts=optional, leveloffset=+1]

If you are interested in including the generated documentation for the config group, you can use the include statement below

include::{generated-dir}/config/hyphenated-config-group-class-name-with-runtime-or-deployment-namespace-replaced-by-config-group-namespace.adoc[opts=optional, leveloffset=+1]

For example, the io.quarkus.vertx.http.runtime.FormAuthConfig configuration group will be generated in a file named quarkus-vertx-http-config-group-form-auth-config.adoc.

A few recommendations:

  • opts=optional is mandatory as we don’t want the build to fail if only part of the configuration documentation has been generated

  • The documentation is generated with a title level of 2 (i.e. ==).You usually need to adjust it.It can be done with leveloffset=+N.

It is not recommended to include the whole configuration documentation in the middle of your guide as it’s heavy.If you have an application.properties extract with your configuration, just do as follows.

First, include a tip just below your application.properties extract:

[TIP]
For more information about the extension configuration please refer to the <<configuration-reference, Configuration Reference>>.

Then, at the end of your documentation, include the extensive documentation:

[[configuration-reference]]
== Configuration Reference

include::{generated-dir}/config/quarkus-your-extension.adoc[opts=optional, leveloffset=+1]

Finally, generate the documentation and check it out.