Web on Reactive Stack

We touched on “non-blocking” and “functional” but what does reactive mean?

The term, “reactive,” refers to programming models that are built around reacting to change — network components reacting to I/O events, UI controllers reacting to mouse events, and others. In that sense, non-blocking is reactive, because, instead of being blocked, we are now in the mode of reacting to notifications as operations complete or data becomes available.

There is also another important mechanism that we on the Spring team associate with “reactive” and that is non-blocking back pressure. In synchronous, imperative code, blocking calls serve as a natural form of back pressure that forces the caller to wait. In non-blocking code, it becomes important to control the rate of events so that a fast producer does not overwhelm its destination.

Reactive Streams is a small spec (also adopted in Java 9) that defines the interaction between asynchronous components with back pressure. For example a data repository (acting as Publisher) can produce data that an HTTP server (acting as Subscriber) can then write to the response. The main purpose of Reactive Streams is to let the subscriber to control how quickly or how slowly the publisher produces data.

Reactive Streams plays an important role for interoperability. It is of interest to libraries and infrastructure components but less useful as an application API, because it is too low-level. Applications need a higher-level and richer, functional API to compose async logic — similar to the Java 8 Stream API but not only for collections. This is the role that reactive libraries play.

Reactor is the reactive library of choice for Spring WebFlux. It provides the Mono and Flux API types to work on data sequences of 0..1 (Mono) and 0..N (Flux) through a rich set of operators aligned with the ReactiveX vocabulary of operators. Reactor is a Reactive Streams library and, therefore, all of its operators support non-blocking back pressure. Reactor has a strong focus on server-side Java. It is developed in close collaboration with Spring.

WebFlux requires Reactor as a core dependency but it is interoperable with other reactive libraries via Reactive Streams. As a general rule, a WebFlux API accepts a plain Publisher as input, adapts it to a Reactor type internally, uses that, and returns either a Flux or a Mono as output. So, you can pass any Publisher as input and you can apply operations on the output, but you need to adapt the output for use with another reactive library. Whenever feasible (for example, annotated controllers), WebFlux adapts transparently to the use of RxJava or another reactive library. See Reactive Libraries for more details.

The spring-web module contains the reactive foundation that underlies Spring WebFlux, including HTTP abstractions, Reactive Streams adapters for supported servers, codecs, and a core WebHandler API comparable to the Servlet API but with non-blocking contracts.

On that foundation, Spring WebFlux provides a choice of two programming models:

Spring MVC or WebFlux?

A natural question to ask but one that sets up an unsound dichotomy. Actually, both work together to expand the range of available options. The two are designed for continuity and consistency with each other, they are available side by side, and feedback from each side benefits both sides. The following diagram shows how the two relate, what they have in common, and what each supports uniquely:

We suggest that you consider the following specific points:

Spring WebFlux is supported on Tomcat, Jetty, Servlet 3.1+ containers, as well as on non-Servlet runtimes such as Netty and Undertow. All servers are adapted to a low-level, common API so that higher-level programming models can be supported across servers.

Spring WebFlux does not have built-in support to start or stop a server. However, it is easy to assemble an application from Spring configuration and WebFlux infrastructure and run it with a few lines of code.

Spring Boot has a WebFlux starter that automates these steps. By default, the starter uses Netty, but it is easy to switch to Tomcat, Jetty, or Undertow by changing your Maven or Gradle dependencies. Spring Boot defaults to Netty, because it is more widely used in the asynchronous, non-blocking space and lets a client and a server share resources.

Tomcat and Jetty can be used with both Spring MVC and WebFlux. Keep in mind, however, that the way they are used is very different. Spring MVC relies on Servlet blocking I/O and lets applications use the Servlet API directly if they need to. Spring WebFlux relies on Servlet 3.1 non-blocking I/O and uses the Servlet API behind a low-level adapter and not exposed for direct use.

For Undertow, Spring WebFlux uses Undertow APIs directly without the Servlet API.

Performance has many characteristics and meanings. Reactive and non-blocking generally do not make applications run faster. They can, in some cases, (for example, if using the WebClient to execute remote calls in parallel). On the whole, it requires more work to do things the non-blocking way and that can increase slightly the required processing time.

The key expected benefit of reactive and non-blocking is the ability to scale with a small, fixed number of threads and less memory. That makes applications more resilient under load, because they scale in a more predictable way. In order to observe those benefits, however, you need to have some latency (including a mix of slow and unpredictable network I/O). That is where the reactive stack begins to show its strengths, and the differences can be dramatic.

Both Spring MVC and Spring WebFlux support annotated controllers, but there is a key difference in the concurrency model and the default assumptions for blocking and threads.

In Spring MVC (and servlet applications in general), it is assumed that applications can block the current thread, (for example, for remote calls), and, for this reason, servlet containers use a large thread pool to absorb potential blocking during request handling.

In Spring WebFlux (and non-blocking servers in general), it is assumed that applications do not block, and, therefore, non-blocking servers use a small, fixed-size thread pool (event loop workers) to handle requests.

What if you do need to use a blocking library? Both Reactor and RxJava provide the publishOn operator to continue processing on a different thread. That means there is an easy escape hatch. Keep in mind, however, that blocking APIs are not a good fit for this concurrency model.

In Reactor and RxJava, you declare logic through operators, and, at runtime, a reactive pipeline is formed where data is processed sequentially, in distinct stages. A key benefit of this is that it frees applications from having to protect mutable state because application code within that pipeline is never invoked concurrently.

What threads should you expect to see on a server running with Spring WebFlux?

The Spring Framework does not provide support for starting and stopping servers. To configure the threading model for a server, you need to use server-specific configuration APIs, or, if you use Spring Boot, check the Spring Boot configuration options for each server. You can configure the WebClient directly. For all other libraries, see their respective documentation.

HttpHandler is a simple contract with a single method to handle a request and response. It is intentionally minimal, and its main, and only purpose is to be a minimal abstraction over different HTTP server APIs.

The following table describes the supported server APIs:

The following table describes server dependencies (also see supported versions):

The code snippets below show using the HttpHandler adapters with each server API:

Reactor Netty

Undertow

Tomcat

Jetty

Servlet 3.1+ Container

To deploy as a WAR to any Servlet 3.1+ container, you can extend and include AbstractReactiveWebInitializer in the WAR. That class wraps an HttpHandler with ServletHttpHandlerAdapter and registers that as a Servlet.

The org.springframework.web.server package builds on the HttpHandler contract to provide a general-purpose web API for processing requests through a chain of multiple WebExceptionHandler, multiple WebFilter, and a single WebHandler component. The chain can be put together with WebHttpHandlerBuilder by simply pointing to a Spring ApplicationContext where components are auto-detected, and/or by registering components with the builder.

While HttpHandler has a simple goal to abstract the use of different HTTP servers, the WebHandler API aims to provide a broader set of features commonly used in web applications such as:

Web MVC

In the WebHandler API, you can use a WebFilter to apply interception-style logic before and after the rest of the processing chain of filters and the target WebHandler. When using the WebFlux Config, registering a WebFilter is as simple as declaring it as a Spring bean and (optionally) expressing precedence by using @Order on the bean declaration or by implementing Ordered.

Web MVC

In the WebHandler API, you can use a WebExceptionHandler to handle exceptions from the chain of WebFilter instances and the target WebHandler. When using the WebFlux Config, registering a WebExceptionHandler is as simple as declaring it as a Spring bean and (optionally) expressing precedence by using @Order on the bean declaration or by implementing Ordered.

The following table describes the available WebExceptionHandler implementations:

Web MVC

The spring-web and spring-core modules provide support for serializing and deserializing byte content to and from higher level objects through non-blocking I/O with Reactive Streams back pressure. The following describes this support:

The spring-core module provides byte[], ByteBuffer, DataBuffer, Resource, and String encoder and decoder implementations. The spring-web module provides Jackson JSON, Jackson Smile, JAXB2, Protocol Buffers and other encoders and decoders along with web-only HTTP message reader and writer implementations for form data, multipart content, server-sent events, and others.

ClientCodecConfigurer and ServerCodecConfigurer are typically used to configure and customize the codecs to use in an application. See the section on configuring HTTP message codecs.

Web MVC

DEBUG level logging in Spring WebFlux is designed to be compact, minimal, and human-friendly. It focuses on high value bits of information that are useful over and over again vs others that are useful only when debugging a specific issue.

TRACE level logging generally follows the same principles as DEBUG (and for example also should not be a firehose) but can be used for debugging any issue. In addition some log messages may show a different level of detail at TRACE vs DEBUG.

Good logging comes from the experience of using the logs. If you spot anything that does not meet the stated goals, please let us know.

Web MVC

The DispatcherHandler delegates to special beans to process requests and render the appropriate responses. By “special beans,” we mean Spring-managed Object instances that implement WebFlux framework contracts. Those usually come with built-in contracts, but you can customize their properties, extend them, or replace them.

The following table lists the special beans detected by the DispatcherHandler. Note that there are also some other beans detected at a lower level (see Special bean types in the Web Handler API).

Web MVC

Applications can declare the infrastructure beans (listed under Web Handler API and DispatcherHandler) that are required to process requests. However, in most cases, the WebFlux Config is the best starting point. It declares the required beans and provides a higher-level configuration callback API to customize it.

Web MVC

DispatcherHandler processes requests as follows:

The return value from the invocation of a handler, through a HandlerAdapter, is wrapped as a HandlerResult, along with some additional context, and passed to the first HandlerResultHandler that claims support for it. The following table shows the available HandlerResultHandler implementations, all of which are declared in the WebFlux Config:

Web MVC

The HandlerResult returned from a HandlerAdapter can expose a function for error handling based on some handler-specific mechanism. This error function is called if:

The error function can change the response (for example, to an error status), as long as an error signal occurs before the reactive type returned from the handler produces any data items.

This is how @ExceptionHandler methods in @Controller classes are supported. By contrast, support for the same in Spring MVC is built on a HandlerExceptionResolver. This generally should not matter. However, keep in mind that, in WebFlux, you cannot use a @ControllerAdvice to handle exceptions that occur before a handler is chosen.

See also Managing Exceptions in the “Annotated Controller” section or Exceptions in the WebHandler API section.

Web MVC

View resolution enables rendering to a browser with an HTML template and a model without tying you to a specific view technology. In Spring WebFlux, view resolution is supported through a dedicated HandlerResultHandler that uses ViewResolver instances to map a String (representing a logical view name) to a View instance. The View is then used to render the response.

Web MVC

You can define controller beans by using a standard Spring bean definition. The @Controller stereotype allows for auto-detection and is aligned with Spring general support for detecting @Component classes in the classpath and auto-registering bean definitions for them. It also acts as a stereotype for the annotated class, indicating its role as a web component.

To enable auto-detection of such @Controller beans, you can add component scanning to your Java configuration, as the following example shows:

@RestController is a composed annotation that is itself meta-annotated with @Controller and @ResponseBody, indicating a controller whose every method inherits the type-level @ResponseBody annotation and, therefore, writes directly to the response body versus view resolution and rendering with an HTML template.

Web MVC

The @RequestMapping annotation is used to map requests to controllers methods. It has various attributes to match by URL, HTTP method, request parameters, headers, and media types. You can use it at the class level to express shared mappings or at the method level to narrow down to a specific endpoint mapping.

There are also HTTP method specific shortcut variants of @RequestMapping:

The preceding annotations are Custom Annotations that are provided because, arguably, most controller methods should be mapped to a specific HTTP method versus using @RequestMapping, which, by default, matches to all HTTP methods. At the same time, a @RequestMapping is still needed at the class level to express shared mappings.

The following example uses type and method level mappings:

Web MVC

@RequestMapping handler methods have a flexible signature and can choose from a range of supported controller method arguments and return values.

Web MVC

You can use the @ModelAttribute annotation:

This section discusses @ModelAttribute methods, or the second item from the preceding list. A controller can have any number of @ModelAttribute methods. All such methods are invoked before @RequestMapping methods in the same controller. A @ModelAttribute method can also be shared across controllers through @ControllerAdvice. See the section on Controller Advice for more details.

@ModelAttribute methods have flexible method signatures. They support many of the same arguments as @RequestMapping methods (except for @ModelAttribute itself and anything related to the request body).

The following example uses a @ModelAttribute method:

The following example adds one attribute only:

Spring WebFlux, unlike Spring MVC, explicitly supports reactive types in the model (for example, Mono<Account> or io.reactivex.Single<Account>). Such asynchronous model attributes can be transparently resolved (and the model updated) to their actual values at the time of @RequestMapping invocation, provided a @ModelAttribute argument is declared without a wrapper, as the following example shows:

In addition, any model attributes that have a reactive type wrapper are resolved to their actual values (and the model updated) just prior to view rendering.

You can also use @ModelAttribute as a method-level annotation on @RequestMapping methods, in which case the return value of the @RequestMapping method is interpreted as a model attribute. This is typically not required, as it is the default behavior in HTML controllers, unless the return value is a String that would otherwise be interpreted as a view name. @ModelAttribute can also help to customize the model attribute name, as the following example shows:

Web MVC

@Controller or @ControllerAdvice classes can have @InitBinder methods, to initialize instances of WebDataBinder. Those, in turn, are used to:

@InitBinder methods can register controller-specific java.bean.PropertyEditor or Spring Converter and Formatter components. In addition, you can use the WebFlux Java configuration to register Converter and Formatter types in a globally shared FormattingConversionService.

@InitBinder methods support many of the same arguments that @RequestMapping methods do, except for @ModelAttribute (command object) arguments. Typically, they are declared with a WebDataBinder argument, for registrations, and a void return value. The following example uses the @InitBinder annotation:

Alternatively, when using a Formatter-based setup through a shared FormattingConversionService, you could re-use the same approach and register controller-specific Formatter instances, as the following example shows:

Web MVC

@Controller and @ControllerAdvice classes can have @ExceptionHandler methods to handle exceptions from controller methods. The following example includes such a handler method:

The exception can match against a top-level exception being propagated (that is, a direct IOException being thrown) or against the immediate cause within a top-level wrapper exception (for example, an IOException wrapped inside an IllegalStateException).

For matching exception types, preferably declare the target exception as a method argument, as shown in the preceding example. Alternatively, the annotation declaration can narrow the exception types to match. We generally recommend being as specific as possible in the argument signature and to declare your primary root exception mappings on a @ControllerAdvice prioritized with a corresponding order. See the MVC section for details.

Support for @ExceptionHandler methods in Spring WebFlux is provided by the HandlerAdapter for @RequestMapping methods. See DispatcherHandler for more detail.

Web MVC

Typically, the @ExceptionHandler, @InitBinder, and @ModelAttribute methods apply within the @Controller class (or class hierarchy) in which they are declared. If you want such methods to apply more globally (across controllers), you can declare them in a class annotated with @ControllerAdvice or @RestControllerAdvice.

@ControllerAdvice is annotated with @Component, which means that such classes can be registered as Spring beans through component scanning. @RestControllerAdvice is a composed annotation that is annotated with both @ControllerAdvice and @ResponseBody, which essentially means @ExceptionHandler methods are rendered to the response body through message conversion (versus view resolution or template rendering).

On startup, the infrastructure classes for @RequestMapping and @ExceptionHandler methods detect Spring beans annotated with @ControllerAdvice and then apply their methods at runtime. Global @ExceptionHandler methods (from a @ControllerAdvice) are applied after local ones (from the @Controller). By contrast, global @ModelAttribute and @InitBinder methods are applied before local ones.

By default, @ControllerAdvice methods apply to every request (that is, all controllers), but you can narrow that down to a subset of controllers by using attributes on the annotation, as the following example shows:

The selectors in the preceding example are evaluated at runtime and may negatively impact performance if used extensively. See the @ControllerAdvice javadoc for more details.

Web MVC

In WebFlux.fn, an HTTP request is handled with a HandlerFunction: a function that takes ServerRequest and returns a delayed ServerResponse (i.e. Mono<ServerResponse>). Both the request as the response object have immutable contracts that offer JDK 8-friendly access to the HTTP request and response. HandlerFunction is the equivalent of the body of a @RequestMapping method in the annotation-based programming model.

Incoming requests are routed to a handler function with a RouterFunction: a function that takes ServerRequest and returns a delayed HandlerFunction (i.e. Mono<HandlerFunction>). When the router function matches, a handler function is returned; otherwise an empty Mono. RouterFunction is the equivalent of a @RequestMapping annotation, but with the major difference that router functions provide not just data, but also behavior.

RouterFunctions.route() provides a router builder that facilitates the creation of routers, as the following example shows:

One way to run a RouterFunction is to turn it into an HttpHandler and install it through one of the built-in server adapters:

Most applications can run through the WebFlux Java configuration, see Running a Server.

Web MVC

ServerRequest and ServerResponse are immutable interfaces that offer JDK 8-friendly access to the HTTP request and response. Both request and response provide Reactive Streams back pressure against the body streams. The request body is represented with a Reactor Flux or Mono. The response body is represented with any Reactive Streams Publisher, including Flux and Mono. For more on that, see Reactive Libraries.

Web MVC

Router functions are used to route the requests to the corresponding HandlerFunction. Typically, you do not write router functions yourself, but rather use a method on the RouterFunctions utility class to create one. RouterFunctions.route() (no parameters) provides you with a fluent builder for creating a router function, whereas RouterFunctions.route(RequestPredicate, HandlerFunction) offers a direct way to create a router.

Generally, it is recommended to use the route() builder, as it provides convenient short-cuts for typical mapping scenarios without requiring hard-to-discover static imports. For instance, the router function builder offers the method GET(String, HandlerFunction) to create a mapping for GET requests; and POST(String, HandlerFunction) for POSTs.

Besides HTTP method-based mapping, the route builder offers a way to introduce additional predicates when mapping to requests. For each HTTP method there is an overloaded variant that takes a RequestPredicate as a parameter, though which additional constraints can be expressed.

Web MVC

How do you run a router function in an HTTP server? A simple option is to convert a router function to an HttpHandler by using one of the following:

You can then use the returned HttpHandler with a number of server adapters by following HttpHandler for server-specific instructions.

A more typical option, also used by Spring Boot, is to run with a DispatcherHandler-based setup through the WebFlux Config, which uses Spring configuration to declare the components required to process requests. The WebFlux Java configuration declares the following infrastructure components to support functional endpoints:

The preceding components let functional endpoints fit within the DispatcherHandler request processing lifecycle and also (potentially) run side by side with annotated controllers, if any are declared. It is also how functional endpoints are enabled by the Spring Boot WebFlux starter.

The following example shows a WebFlux Java configuration (see DispatcherHandler for how to run it):

Web MVC

You can filter handler functions by using the before, after, or filter methods on the routing function builder. With annotations, you can achieve similar functionality by using @ControllerAdvice, a ServletFilter, or both. The filter will apply to all routes that are built by the builder. This means that filters defined in nested routes do not apply to "top-level" routes. For instance, consider the following example:

The filter method on the router builder takes a HandlerFilterFunction: a function that takes a ServerRequest and HandlerFunction and returns a ServerResponse. The handler function parameter represents the next element in the chain. This is typically the handler that is routed to, but it can also be another filter if multiple are applied.

Now we can add a simple security filter to our route, assuming that we have a SecurityManager that can determine whether a particular path is allowed. The following example shows how to do so:

The preceding example demonstrates that invoking the next.handle(ServerRequest) is optional. We allow only the handler function to be executed when access is allowed.

Besides using the filter method on the router function builder, it is possible to apply a filter to an existing router function via RouterFunction.filter(HandlerFilterFunction).

Spring MVC and Spring WebFlux

UriComponentsBuilder helps to build URI’s from URI templates with variables, as the following example shows:

The preceding example can be consolidated into one chain and shortened with buildAndExpand, as the following example shows:

You can shorten it further by going directly to a URI (which implies encoding), as the following example shows:

You shorter it further still with a full URI template, as the following example shows:

Spring MVC and Spring WebFlux

UriComponentsBuilder implements UriBuilder. You can create a UriBuilder, in turn, with a UriBuilderFactory. Together, UriBuilderFactory and UriBuilder provide a pluggable mechanism to build URIs from URI templates, based on shared configuration, such as a base URL, encoding preferences, and other details.

You can configure RestTemplate and WebClient with a UriBuilderFactory to customize the preparation of URIs. DefaultUriBuilderFactory is a default implementation of UriBuilderFactory that uses UriComponentsBuilder internally and exposes shared configuration options.

The following example shows how to configure a RestTemplate:

The following example configures a WebClient:

In addition, you can also use DefaultUriBuilderFactory directly. It is similar to using UriComponentsBuilder but, instead of static factory methods, it is an actual instance that holds configuration and preferences, as the following example shows:

Spring MVC and Spring WebFlux

UriComponentsBuilder exposes encoding options at two levels:

Both options replace non-ASCII and illegal characters with escaped octets. However, the first option also replaces characters with reserved meaning that appear in URI variables.

For most cases, the first option is likely to give the expected result, because it treats URI variables as opaque data to be fully encoded, while option 2 is useful only if URI variables intentionally contain reserved characters.

The following example uses the first option:

You can shorten the preceding example by going directly to the URI (which implies encoding), as the following example shows:

You can shorten it further still with a full URI template, as the following example shows:

The WebClient and the RestTemplate expand and encode URI templates internally through the UriBuilderFactory strategy. Both can be configured with a custom strategy. as the following example shows:

The DefaultUriBuilderFactory implementation uses UriComponentsBuilder internally to expand and encode URI templates. As a factory, it provides a single place to configure the approach to encoding, based on one of the below encoding modes:

The RestTemplate is set to EncodingMode.URI_COMPONENTS for historic reasons and for backwards compatibility. The WebClient relies on the default value in DefaultUriBuilderFactory, which was changed from EncodingMode.URI_COMPONENTS in 5.0.x to EncodingMode.TEMPLATE_AND_VALUES in 5.1.

Web MVC

For security reasons, browsers prohibit AJAX calls to resources outside the current origin. For example, you could have your bank account in one tab and evil.com in another. Scripts from evil.com should not be able to make AJAX requests to your bank API with your credentials — for example, withdrawing money from your account!

Cross-Origin Resource Sharing (CORS) is a W3C specification implemented by most browsers that lets you specify what kind of cross-domain requests are authorized, rather than using less secure and less powerful workarounds based on IFRAME or JSONP.

Web MVC

The CORS specification distinguishes between preflight, simple, and actual requests. To learn how CORS works, you can read this article, among many others, or see the specification for more details.

Spring WebFlux HandlerMapping implementations provide built-in support for CORS. After successfully mapping a request to a handler, a HandlerMapping checks the CORS configuration for the given request and handler and takes further actions. Preflight requests are handled directly, while simple and actual CORS requests are intercepted, validated, and have the required CORS response headers set.

In order to enable cross-origin requests (that is, the Origin header is present and differs from the host of the request), you need to have some explicitly declared CORS configuration. If no matching CORS configuration is found, preflight requests are rejected. No CORS headers are added to the responses of simple and actual CORS requests and, consequently, browsers reject them.

Each HandlerMapping can be configured individually with URL pattern-based CorsConfiguration mappings. In most cases, applications use the WebFlux Java configuration to declare such mappings, which results in a single, global map passed to all HadlerMappping implementations.

You can combine global CORS configuration at the HandlerMapping level with more fine-grained, handler-level CORS configuration. For example, annotated controllers can use class- or method-level @CrossOrigin annotations (other handlers can implement CorsConfigurationSource).

The rules for combining global and local configuration are generally additive — for example, all global and all local origins. For those attributes where only a single value can be accepted, such as allowCredentials and maxAge, the local overrides the global value. See CorsConfiguration#combine(CorsConfiguration) for more details.

Web MVC

The @CrossOrigin annotation enables cross-origin requests on annotated controller methods, as the following example shows:

By default, @CrossOrigin allows:

allowedCredentials is not enabled by default, since that establishes a trust level that exposes sensitive user-specific information (such as cookies and CSRF tokens) and should be used only where appropriate.

maxAge is set to 30 minutes.

@CrossOrigin is supported at the class level, too, and inherited by all methods. The following example specifies a certain domain and sets maxAge to an hour:

You can use @CrossOrigin at both the class and the method level, as the following example shows:

Web MVC

In addition to fine-grained, controller method-level configuration, you probably want to define some global CORS configuration, too. You can set URL-based CorsConfiguration mappings individually on any HandlerMapping. Most applications, however, use the WebFlux Java configuration to do that.

By default global configuration enables the following:

allowedCredentials is not enabled by default, since that establishes a trust level that exposes sensitive user-specific information( such as cookies and CSRF tokens) and should be used only where appropriate.

maxAge is set to 30 minutes.

To enable CORS in the WebFlux Java configuration, you can use the CorsRegistry callback, as the following example shows:

Web MVC

You can apply CORS support through the built-in CorsWebFilter, which is a good fit with functional endpoints.

To configure the filter, you can declare a CorsWebFilter bean and pass a CorsConfigurationSource to its constructor, as the following example shows:

Web MVC

Thymeleaf is a modern server-side Java template engine that emphasizes natural HTML templates that can be previewed in a browser by double-clicking, which is very helpful for independent work on UI templates (for example, by a designer) without the need for a running server. Thymeleaf offers an extensive set of features, and it is actively developed and maintained. For a more complete introduction, see the Thymeleaf project home page.

The Thymeleaf integration with Spring WebFlux is managed by the Thymeleaf project. The configuration involves a few bean declarations, such as SpringResourceTemplateResolver, SpringWebFluxTemplateEngine, and ThymeleafReactiveViewResolver. For more details, see Thymeleaf+Spring and the WebFlux integration announcement.

Web MVC

Apache FreeMarker is a template engine for generating any kind of text output from HTML to email and others. The Spring Framework has built-in integration for using Spring WebFlux with FreeMarker templates.

Web MVC

The Spring Framework has a built-in integration for using Spring WebFlux with any templating library that can run on top of the JSR-223 Java scripting engine. The following table shows the templating libraries that we have tested on different script engines:

Web MVC

For Content Negotiation purposes, it is useful to be able to alternate between rendering a model with an HTML template or as other formats (such as JSON or XML), depending on the content type requested by the client. To support doing so, Spring WebFlux provides the HttpMessageWriterView, which you can use to plug in any of the available Codecs from spring-web, such as Jackson2JsonEncoder, Jackson2SmileEncoder, or Jaxb2XmlEncoder.

Unlike other view technologies, HttpMessageWriterView does not require a ViewResolver but is instead configured as a default view. You can configure one or more such default views, wrapping different HttpMessageWriter instances or Encoder instances. The one that matches the requested content type is used at runtime.

In most cases, a model contains multiple attributes. To determine which one to serialize, you can configure HttpMessageWriterView with the name of the model attribute to use for rendering. If the model contains only one attribute, that one is used.

Web MVC

CacheControl provides support for configuring settings related to the Cache-Control header and is accepted as an argument in a number of places:

While RFC 7234 describes all possible directives for the Cache-Control response header, the CacheControl type takes a use case-oriented approach that focuses on the common scenarios, as the following example shows:

Web MVC

Controllers can add explicit support for HTTP caching. We recommend doing so, since the lastModified or ETag value for a resource needs to be calculated before it can be compared against conditional request headers. A controller can add an ETag and Cache-Control settings to a ResponseEntity, as the following example shows:

The preceding example sends a 304 (NOT_MODIFIED) response with an empty body if the comparison to the conditional request headers indicates the content has not changed. Otherwise, the ETag and Cache-Control headers are added to the response.

You can also make the check against conditional request headers in the controller, as the following example shows:

There are three variants for checking conditional requests against eTag values, lastModified values, or both. For conditional GET and HEAD requests, you can set the response to 304 (NOT_MODIFIED). For conditional POST, PUT, and DELETE, you can instead set the response to 409 (PRECONDITION_FAILED) to prevent concurrent modification.

Web MVC

You should serve static resources with a Cache-Control and conditional response headers for optimal performance. See the section on configuring Static Resources.

Web MVC

You can use the @EnableWebFlux annotation in your Java config, as the following example shows:

The preceding example registers a number of Spring WebFlux infrastructure beans and adapts to dependencies available on the classpath — for JSON, XML, and others.

Web MVC

In your Java configuration, you can implement the WebFluxConfigurer interface, as the following example shows:

Web MVC

By default, formatters for Number and Date types are installed, including support for the @NumberFormat and @DateTimeFormat annotations. Full support for the Joda-Time formatting library is also installed if Joda-Time is present on the classpath.

The following example shows how to register custom formatters and converters:

Web MVC

By default, if Bean Validation is present on the classpath (for example, the Hibernate Validator), the LocalValidatorFactoryBean is registered as a global validator for use with @Valid and Validated on @Controller method arguments.

In your Java configuration, you can customize the global Validator instance, as the following example shows:

Note that you can also register Validator implementations locally, as the following example shows:

Web MVC

You can configure how Spring WebFlux determines the requested media types for @Controller instances from the request. By default, only the Accept header is checked, but you can also enable a query parameter-based strategy.

The following example shows how to customize the requested content type resolution:

Web MVC

The following example shows how to customize how the request and response body are read and written:

ServerCodecConfigurer provides a set of default readers and writers. You can use it to add more readers and writers, customize the default ones, or replace the default ones completely.

For Jackson JSON and XML, consider using Jackson2ObjectMapperBuilder, which customizes Jackson’s default properties with the following ones:

It also automatically registers the following well-known modules if they are detected on the classpath:

Web MVC

The following example shows how to configure view resolution:

The ViewResolverRegistry has shortcuts for view technologies with which the Spring Framework integrates. The following example uses FreeMarker (which also requires configuring the underlying FreeMarker view technology):

You can also plug in any ViewResolver implementation, as the following example shows:

To support Content Negotiation and rendering other formats through view resolution (besides HTML), you can configure one or more default views based on the HttpMessageWriterView implementation, which accepts any of the available Codecs from spring-web. The following example shows how to do so:

See View Technologies for more on the view technologies that are integrated with Spring WebFlux.

Web MVC

This option provides a convenient way to serve static resources from a list of Resource-based locations.

In the next example, given a request that starts with /resources, the relative path is used to find and serve static resources relative to /static on the classpath. Resources are served with a one-year future expiration to ensure maximum use of the browser cache and a reduction in HTTP requests made by the browser. The Last-Modified header is also evaluated and, if present, a 304 status code is returned. The following list shows the example:

The resource handler also supports a chain of ResourceResolver implementations and ResourceTransformer implementations, which can be used to create a toolchain for working with optimized resources.

You can use the VersionResourceResolver for versioned resource URLs based on an MD5 hash computed from the content, a fixed application version, or other information. A ContentVersionStrategy (MD5 hash) is a good choice with some notable exceptions (such as JavaScript resources used with a module loader).

The following example shows how to use VersionResourceResolver in your Java configuration:

You can use ResourceUrlProvider to rewrite URLs and apply the full chain of resolvers and transformers (for example, to insert versions). The WebFlux configuration provides a ResourceUrlProvider so that it can be injected into others.

Unlike Spring MVC, at present, in WebFlux, there is no way to transparently rewrite static resource URLs, since there are no view technologies that can make use of a non-blocking chain of resolvers and transformers. When serving only local resources, the workaround is to use ResourceUrlProvider directly (for example, through a custom element) and block.

Note that, when using both EncodedResourceResolver (for example, Gzip, Brotli encoded) and VersionedResourceResolver, they must be registered in that order, to ensure content-based versions are always computed reliably based on the unencoded file.

WebJars are also supported through the WebJarsResourceResolver which is automatically registered when the org.webjars:webjars-locator-core library is present on the classpath. The resolver can re-write URLs to include the version of the jar and can also match against incoming URLs without versions — for example, from /jquery/jquery.min.js to /jquery/1.2.0/jquery.min.js.

Web MVC

You can customize options related to path matching. For details on the individual options, see the PathMatchConfigurer javadoc. The following example shows how to use PathMatchConfigurer:

Web MVC

@EnableWebFlux imports DelegatingWebFluxConfiguration that:

For advanced mode, you can remove @EnableWebFlux and extend directly from DelegatingWebFluxConfiguration instead of implementing WebFluxConfigurer, as the following example shows:

You can keep existing methods in WebConfig, but you can now also override bean declarations from the base class and still have any number of other WebMvcConfigurer implementations on the classpath.

Spring WebFlux configures limits for buffering data in-memory in codec to avoid application memory issues. By the default this is configured to 256KB and if that’s not enough for your use case, you’ll see the following:

You can configure this limit on all default codecs with the following code sample:

To customize Reactor Netty settings, simple provide a pre-configured HttpClient:

The following example shows how to customize Jetty HttpClient settings:

By default, HttpClient creates its own resources (Executor, ByteBufferPool, Scheduler), which remain active until the process exits or stop() is called.

You can share resources between multiple instances of the Jetty client (and server) and ensure that the resources are shut down when the Spring ApplicationContext is closed by declaring a Spring-managed bean of type JettyResourceFactory, as the following example shows:

To send form data, you can provide a MultiValueMap<String, String> as the body. Note that the content is automatically set to application/x-www-form-urlencoded by the FormHttpMessageWriter. The following example shows how to use MultiValueMap<String, String>:

You can also supply form data in-line by using BodyInserters, as the following example shows:

To send multipart data, you need to provide a MultiValueMap<String, ?> whose values are either Object instances that represent part content or HttpEntity instances that represent the content and headers for a part. MultipartBodyBuilder provides a convenient API to prepare a multipart request. The following example shows how to create a MultiValueMap<String, ?>:

In most cases, you do not have to specify the Content-Type for each part. The content type is determined automatically based on the HttpMessageWriter chosen to serialize it or, in the case of a Resource, based on the file extension. If necessary, you can explicitly provide the MediaType to use for each part through one of the overloaded builder part methods.

Once a MultiValueMap is prepared, the easiest way to pass it to the WebClient is through the body method, as the following example shows:

If the MultiValueMap contains at least one non-String value, which could also represent regular form data (that is, application/x-www-form-urlencoded), you need not set the Content-Type to multipart/form-data. This is always the case when using MultipartBodyBuilder, which ensures an HttpEntity wrapper.

As an alternative to MultipartBodyBuilder, you can also provide multipart content, inline-style, through the built-in BodyInserters, as the following example shows:

Even though WebSocket is designed to be HTTP-compatible and starts with an HTTP request, it is important to understand that the two protocols lead to very different architectures and application programming models.

In HTTP and REST, an application is modeled as many URLs. To interact with the application, clients access those URLs, request-response style. Servers route requests to the appropriate handler based on the HTTP URL, method, and headers.

By contrast, in WebSockets, there is usually only one URL for the initial connect. Subsequently, all application messages flow on that same TCP connection. This points to an entirely different asynchronous, event-driven, messaging architecture.

WebSocket is also a low-level transport protocol, which, unlike HTTP, does not prescribe any semantics to the content of messages. That means that there is no way to route or process a message unless the client and the server agree on message semantics.

WebSocket clients and servers can negotiate the use of a higher-level, messaging protocol (for example, STOMP), through the Sec-WebSocket-Protocol header on the HTTP handshake request. In the absence of that, they need to come up with their own conventions.

WebSockets can make a web page be dynamic and interactive. However, in many cases, a combination of Ajax and HTTP streaming or long polling can provide a simple and effective solution.

For example, news, mail, and social feeds need to update dynamically, but it may be perfectly okay to do so every few minutes. Collaboration, games, and financial apps, on the other hand, need to be much closer to real-time.

Latency alone is not a deciding factor. If the volume of messages is relatively low (for example, monitoring network failures) HTTP streaming or polling can provide an effective solution. It is the combination of low latency, high frequency, and high volume that make the best case for the use of WebSocket.

Keep in mind also that over the Internet, restrictive proxies that are outside of your control may preclude WebSocket interactions, either because they are not configured to pass on the Upgrade header or because they close long-lived connections that appear idle. This means that the use of WebSocket for internal applications within the firewall is a more straightforward decision than it is for public facing applications.

Same as in the Servlet stack

To create a WebSocket server, you can first create a WebSocketHandler. The following example shows how to do so:

Then you can map it to a URL and add a WebSocketHandlerAdapter, as the following example shows:

The handle method of WebSocketHandler takes WebSocketSession and returns Mono<Void> to indicate when application handling of the session is complete. The session is handled through two streams, one for inbound and one for outbound messages. The following table describes the two methods that handle the streams:

A WebSocketHandler must compose the inbound and outbound streams into a unified flow and return a Mono<Void> that reflects the completion of that flow. Depending on application requirements, the unified flow completes when:

When inbound and outbound message streams are composed together, there is no need to check if the connection is open, since Reactive Streams signals terminate activity. The inbound stream receives a completion or error signal, and the outbound stream receives a cancellation signal.

The most basic implementation of a handler is one that handles the inbound stream. The following example shows such an implementation:

The following implementation combines the inbound and outbound streams:

Inbound and outbound streams can be independent and be joined only for completion, as the following example shows:

DataBuffer is the representation for a byte buffer in WebFlux. The Spring Core part of the reference has more on that in the section on Data Buffers and Codecs. The key point to understand is that on some servers like Netty, byte buffers are pooled and reference counted, and must be released when consumed to avoid memory leaks.

When running on Netty, applications must use DataBufferUtils.retain(dataBuffer) if they wish to hold on input data buffers in order to ensure they are not released, and subsequently use DataBufferUtils.release(dataBuffer) when the buffers are consumed.

Same as in the Servlet stack

WebSocketHandlerAdapter delegates to a WebSocketService. By default, that is an instance of HandshakeWebSocketService, which performs basic checks on the WebSocket request and then uses RequestUpgradeStrategy for the server in use. Currently, there is built-in support for Reactor Netty, Tomcat, Jetty, and Undertow.

HandshakeWebSocketService exposes a sessionAttributePredicate property that allows setting a Predicate<String> to extract attributes from the WebSession and insert them into the attributes of the WebSocketSession.

Same as in the Servlet stack

The RequestUpgradeStrategy for each server exposes WebSocket-related configuration options available for the underlying WebSocket engine. The following example sets WebSocket options when running on Tomcat:

Check the upgrade strategy for your server to see what options are available. Currently, only Tomcat and Jetty expose such options.

Same as in the Servlet stack

The easiest way to configure CORS and restrict access to a WebSocket endpoint is to have your WebSocketHandler implement CorsConfigurationSource and return a CorsConfiguraiton with allowed origins, headers, and other details. If you cannot do that, you can also set the corsConfigurations property on the SimpleUrlHandler to specify CORS settings by URL pattern. If both are specified, they are combined by using the combine method on CorsConfiguration.

Spring WebFlux provides a WebSocketClient abstraction with implementations for Reactor Netty, Tomcat, Jetty, Undertow, and standard Java (that is, JSR-356).

To start a WebSocket session, you can create an instance of the client and use its execute methods:

Some clients, such as Jetty, implement Lifecycle and need to be stopped and started before you can use them. All clients have constructor options related to configuration of the underlying WebSocket client.

One of the benefits of RSocket is that it has well defined behavior on the wire and an easy to read specification along with some protocol extensions. Therefore it is a good idea to read the spec, independent of language implementations and higher level framework APIs. This section provides a succinct overview to establish some context.

Connecting

Initially a client connects to a server via some low level streaming transport such as TCP or WebSocket and sends a SETUP frame to the server to set parameters for the connection.

The server may reject the SETUP frame, but generally after it is sent (for the client) and received (for the server), both sides can begin to make requests, unless SETUP indicates use of leasing semantics to limit the number of requests, in which case both sides must wait for a LEASE frame from the other end to permit making requests.

Making Requests

Once a connection is established, both sides may initiate a request through one of the frames REQUEST_RESPONSE, REQUEST_STREAM, REQUEST_CHANNEL, or REQUEST_FNF. Each of those frames carries one message from the requester to the responder.

The responder may then return PAYLOAD frames with response messages, and in the case of REQUEST_CHANNEL the requester may also send PAYLOAD frames with more request messages.

When a request involves a stream of messages such as Request-Stream and Channel, the responder must respect demand signals from the requester. Demand is expressed as a number of messages. Initial demand is specified in REQUEST_STREAM and REQUEST_CHANNEL frames. Subsequent demand is signaled via REQUEST_N frames.

Each side may also send metadata notifications, via the METADATA_PUSH frame, that do not pertain to any individual request but rather to the connection as a whole.

Message Format

RSocket messages contain data and metadata. Metadata can be used to send a route, a security token, etc. Data and metadata can be formatted differently. Mime types for each are declared in the SETUP frame and apply to all requests on a given connection.

While all messages can have metadata, typically metadata such as a route are per-request and therefore only included in the first message on a request, i.e. with one of the frames REQUEST_RESPONSE, REQUEST_STREAM, REQUEST_CHANNEL, or REQUEST_FNF.

Protocol extensions define common metadata formats for use in applications:

The Java implementation for RSocket is built on Project Reactor. The transports for TCP and WebSocket are built on Reactor Netty. As a Reactive Streams library, Reactor simplifies the job of implementing the protocol. For applications it is a natural fit to use Flux and Mono with declarative operators and transparent back pressure support.

The API in RSocket Java is intentionally minimal and basic. It focuses on protocol features and leaves the application programming model (e.g. RPC codegen vs other) as a higher level, independent concern.

The main contract io.rsocket.RSocket models the four request interaction types with Mono representing a promise for a single message, Flux a stream of messages, and io.rsocket.Payload the actual message with access to data and metadata as byte buffers. The RSocket contract is used symmetrically. For requesting, the application is given an RSocket to perform requests with. For responding, the application implements RSocket to handle requests.

This is not meant to be a thorough introduction. For the most part, Spring applications will not have to use its API directly. However it may be important to see or experiment with RSocket independent of Spring. The RSocket Java repository contains a number of sample apps that demonstrate its API and protocol features.

The spring-messaging module contains the following:

The spring-web module contains Encoder and Decoder implementations such as Jackson CBOR/JSON, and Protobuf that RSocket applications will likely need. It also contains the PathPatternParser that can be plugged in for efficient route matching.

Spring Boot 2.2 supports standing up an RSocket server over TCP or WebSocket, including the option to expose RSocket over WebSocket in a WebFlux server. There is also client support and auto-configuration for an RSocketRequester.Builder and RSocketStrategies. See the RSocket section in the Spring Boot reference for more details.

Spring Security 5.2 provides RSocket support.

Spring Integration 5.2 provides inbound and outbound gateways to interact with RSocket clients and servers. See the Spring Integration Reference Manual for more details.

Spring Cloud Gateway supports RSocket connections.

To obtain an RSocketRequester on the client side requires connecting to a server along with preparing and sending the initial RSocket SETUP frame. RSocketRequester provides a builder for that. Internally uses RSocket Java’s RSocketFactory.

This is the most basic way to connect with default settings:

The above is deferred. To actually connect and use the requester: