A library to handle success and failure without exceptions

Wave goodbye to slow exceptions and embrace clean, efficient error handling by encapsulating operations that may succeed or fail in a type-safe way.

Results in a Nutshell

In Java, methods that can fail typically do so by throwing exceptions. Then, exception-throwing methods are called from inside a try block to handle errors in a separate catch block.

This approach is lengthy, and that's not the only problem — it's also very slow.

Let's now look at how the above code could be refactored if connect() returned a Result object instead of throwing an exception.

In the example above, we used only 4 lines of code to replace the 10 that worked for the first one. But we can effortlessly make it shorter by chaining methods. In fact, since we were returning -1 just to signal that the underlying operation failed, we are better off returning a Result object upstream. This will allow us to compose operations on top of getServerUptime() just like we did with connect().

Ready to Tap into the Power of Results?

Read the guide and transform your error handling today.

Also available as an ebook in multiple formats. Download your free copy now!

TL;DR

Not a fan of reading long docs? No worries! Tune in to Deep Dive, a podcast generated by NetbookLM. In just a few minutes, you'll get the essential details and a fun intro to what this library can do for you!

Result builds upon the familiar concept of Optional, enhancing it with the ability to represent both success and failure states.

By leveraging Results, you can unleash a powerful tool for error handling that goes beyond the capabilities of traditional Optionals, leading to more robust and maintainable Java code.

As we discovered earlier, we can easily determine if a given Result instance is successful or not.

Checking Success

We can use Result::hasSuccess to obtain a boolean value that represents whether a result is successful.

@Test
void testHasSuccess() {
  // Given
  Result<?, ?> result1 = success(1024);
  Result<?, ?> result2 = failure(1024);
  // When
  boolean result1HasSuccess = result1.hasSuccess();
  boolean result2HasSuccess = result2.hasSuccess();
  // Then
  assertTrue(result1HasSuccess);
  assertFalse(result2HasSuccess);
}

Checking Failure

We can also use Result::hasFailure to find out if a result contains a failure value.

@Test
void testHasFailure() {
  // Given
  Result<?, ?> result1 = success(512);
  Result<?, ?> result2 = failure(512);
  // When
  boolean result1HasFailure = result1.hasFailure();
  boolean result2HasFailure = result2.hasFailure();
  // Then
  assertFalse(result1HasFailure);
  assertTrue(result2HasFailure);
}

Conclusion

We discussed how to determine the state of a Result object using hasSuccess and hasFailure. These methods provide a straightforward way to identify the outcome of an operation, helping you make decisions based on the outcome.

In this section, we'll cover foundational use cases, including checking the status of a result, unwrapping the value inside a result, and taking different actions based on success or failure.

These basics will help you handle errors more cleanly and efficiently without cluttering your code with try-catch blocks.

This library adheres to to communicate the backwards compatibility of each version.

The latest releases are available in .

Artifact Coordinates

Add this Maven dependency to your build:

Maven

To use Result, we can add a dependency to our project.

Gradle

We can also add Result as a dependency.

Conclusion

We learned how to add the library to your project using either or . By including the correct dependencies, you're now ready to start leveraging the power of Results in your applications.

Congratulations on reaching the end of this guide! By now, you should have a solid understanding of how to use results in your Java applications effectively. Here's a brief recap of what you've learned:

• Getting Started: You learned how to integrate result objects into your codebase and instantiate new ones.

• Basic Usage: You explored foundational operations like checking statuses, unwrapping values, and executing conditional actions based on result status, enabling you to respond dynamically to success and failure scenarios.

• Advanced Usage: You delved into more sophisticated techniques like screening results to transform successes and failures based on conditions, and leveraging mapping and flat-mapping methods to compose behaviors in a functional style.

For more details on the Result API, you can read the .

Next, we'll introduce additional resources where you can further enhance your understanding and skills. Let's continue expanding your knowledge!

While understanding the basics provides a solid foundation, the true potential of result objects is unlocked through their functional capabilities. Mastering these techniques enables concise and readable error handling by leveraging the power of monadic composition.

This section will guide you through these powerful tools, demonstrating how to manipulate results effectively so you can craft more robust and maintainable Java applications.

In essence, a Result object is just a container that wraps a success or a failure value for us. Therefore, sometimes you are going to want to get that value out of the container.

Unwrapping Success

The most basic way to retrieve the success value wrapped inside a result is by using . This method will return an optional success value, depending on whether the result was actually successful or not.

Unwrapping Failure

Similarly, we can use to obtain the failure value held by a Result object.

Using Alternative Success

We can use to provide an alternative success value that must be returned when the result is unsuccessful.

Mapping Failure

The method is similar to , but it takes a mapping instead of a . The function will receive the failure value to produce the alternative success value.

Streaming Success or Failure

Finally, we can use and to wrap the value held by an instance of Result into a possibly-empty object.

Conclusion

We explored various ways to retrieve values from results. Using these methods you can efficiently access the underlying data within a Result object, whether it's a success or a failure.

Tracking multiple add-on versions for your project can quickly become cumbersome. In that situation, you can use the convenient to centralize and align their versions. This ensures compatibility and simplifies dependency maintenance.

The basic idea is that instead of specifying a version number for each Result library in your project, you can use this BOM to get a complete set of consistent versions.

How to Use this Add-On

Add this Maven dependency to your build:

Maven

To , use the following:

Gradle

To , use the following:

Conclusion

We discussed the benefits of using the Bill of Materials for managing dependencies in your project. With the BOM, you can eliminate the hassle of manually specifying version numbers, ensuring consistency and compatibility across all Result libraries.

This library is licensed under the Apache License, Version 2.0 (the "License"); you may not use it except in compliance with the License.

You may obtain a copy of the License at

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.

See the License for the specific language governing permissions and limitations under the License.

To help you become familiar with this library, you can explore two demo projects that showcase how to handle and serialize Result objects within popular frameworks like and . Each project provides a working example of a "pet store" web service that exposes a REST API for managing pets. They are based on and you can interact with them using .

These projects illustrate how to develop powerful APIs using Result objects. Follow the examples to create resilient web services that elegantly handle success and failure scenarios.

This demo project demonstrates how to handle and serialize Result objects within a Micronaut application. It provides a working example of a "pet store" web service that exposes a REST API for managing pets.

Generating the Project

The project was generated via Micronaut Launch including features: annotation-api, http-client, openapi, serialization-jackson, swagger-ui, toml, and validation.

Adding Serialization Support

Then Micronaut Serialization for Result objects was manually added as a dependency to serialize and deserialize Result objects.

dependencies {
    // ...
    implementation(platform("com.leakyabstractions:result-bom:1.0.0.0"))
    implementation("com.leakyabstractions:result")
    implementation("com.leakyabstractions:result-micronaut-serde")
}

That's all we need to do to make Micronaut treat results as Serdeable.

API Responses

API responses contain a Result field, encapsulating the outcome of the requested operation.

@Serdeable
public class ApiResponse<S> {

  @JsonProperty String version;
  @JsonProperty Instant generatedOn;
  @JsonProperty Result<S, ApiError> result;
}

Results have different success types, depending on the specific endpoint. Failures will be encapsulated as instances of ApiError.

Controllers

Controllers return instances of ApiResponse that will be serialized to JSON by Micronaut:

@Controller
public class PetController {
  // ...
  @Get("/pet")
  ApiResponse<Collection<Pet>> list(@Header("X-Type") RepositoryType type) {
    log.info("List all pets in {} pet store", type);
    return response(locate(type)
        .flatMapSuccess(PetRepository::listPets)
        .ifSuccess(x -> log.info("Listed {} pet(s) in {}", x.size(), type))
        .ifFailure(this::logError));
  }
}

Since failures are expressed as ApiError objects, endpoints invariably return HTTP status 200.

Running the Application

The application can be built and run with Gradle.

./gradlew run

This will start a stand-alone server on port 8080.

Testing the Server

Once started, you can interact with the API.

curl -s -H 'x-type: local' http://localhost:8080/pet/0

You should see a JSON response like this:

{
  "version": "1.0",
  "result": {
    "success":{
      "id": 0,
      "name": "Rocky",
      "status": "AVAILABLE"
    }
  }
}

Using Swagger-UI

You can navigate to http://localhost:8080/ to inspect the API using an interactive UI.

Throughout these guides, we have mentioned that throwing Java exceptions is slow. But... how slow? According to our benchmarks, throwing an exception is several orders of magnitude slower than returning a failed result.

This proves that using exceptional logic just to control normal program flow is a bad idea.

Benchmarking Result Library

This library comes with a set of benchmarks that compare performance when using results versus when using exceptions.

Simple Scenarios

The first scenarios compare the most basic usage: a method that returns a String or fails, depending on a given int parameter:

Using Exceptions

public String usingExceptions(int number) throws SimpleException {
  if (number < 0) {
    throw new SimpleException(number);
  }
  return "ok";
}

Using Results

public Result<String, SimpleFailure> usingResults(int number) {
  if (number < 0) {
    return Results.failure(new SimpleFailure(number));
  }
  return Results.success("ok");
}

Complex Scenarios

The next scenarios do something a little bit more elaborate: a method invokes the previous method to retrieve a String; if successful, then converts it to upper case; otherwise transforms the "simple" error into a "complex" error.

Using Exceptions

public String usingExceptions(int number) throws ComplexException {
  try {
    return simple.usingExceptions(number).toUpperCase();
  } catch (SimpleException e) {
    throw new ComplexException(e);
  }
}

Using Results

public Result<String, ComplexFailure> usingResults(int number) {
  return simple.usingResults(number)
    .map(String::toUpperCase, ComplexFailure::new);
}

Conclusion

We provided insights into the Result library's performance through benchmarking. While our metrics corroborate that most codebases could benefit from using this library instead of throwing exceptions, its main goal is to help promote best practices and implement proper error handling.

The following methods allow you to run inline tests on the wrapped value of a result to dynamically transform a success into a failure or a failure into a success.

Validating Success

The method allows you to transform a success into a failure based on certain conditions. It takes two parameters:

1. A to determine if the success value is acceptable.

2. A mapping that will produce a failure value if the value is deemed unacceptable.

In this example, we use a lambda expression to validate that the success value inside result is even. Since the number is odd, it transforms the result into a failure.

Recovering From Failure

The method allows you to transform a failure into a success based on certain conditions. It also receives two parameters:

1. A to determine if the failure value is recoverable.

2. A mapping that will produce a success value from the acceptable failure value.

In this example, we use method references to check if the failure value equals OK and then transform the result into a success.

Conclusion

We covered how to filter out unwanted success values and accept failure values using and . These methods enable you to refine results based on specific criteria, ensuring that only the relevant values are processed down the line.

Transforming result objects is a key feature that enables you to compose complex operations in a clean and functional style. There are two primary techniques used for these transformations.

Mapping Results

Mapping involves applying a function to the value inside a result to produce a new result object.

Mapping Success Values

We can use Result::mapSuccess to apply a function to the success value of a result, transforming it into a new success value. If the result is a failure, it remains unchanged.

@Test
void testMapSuccess() {
  // Given
  Result<String, ?> result = success("HELLO");
  // When
  Result<Integer, ?> mapped = result.mapSuccess(String::length);
  // Then
  assertEquals(5, mapped.orElse(null));
}

In this example, we wrap a String inside a Result object and invoke mapSuccess to calculate its length and wrap it inside a new Result object.

Mapping Failure Values

Next up, we can use Result::mapFailure to apply a function to the failure value, transforming it into a new one. If the result is a success, it remains unchanged.

@Test
void testMapFailure() {
  // Given
  Result<?, BigDecimal> result = failure(ONE);
  // When
  Result<?, Boolean> mapped = result.mapFailure(TWO::equals);
  // Then
  assertFalse(mapped.getFailure().orElse(null));
}

Here, we invoke mapFailure to transform the failure type of the result from String to Boolean for demonstration purposes.

Mapping Both Success and Failure

The Result::map method simultaneously handles both success and failure cases by applying two separate functions: one for transforming the success value and one for transforming the failure value.

@Test
void testMap() {
  // Given
  Result<String, BigDecimal> result1 = success("HELLO");
  Result<String, BigDecimal> result2 = failure(ONE);
  // When
  Result<Integer, Boolean> mapped1 = result1.map(String::length, TWO::equals);
  Result<Integer, Boolean> mapped2 = result2.map(String::length, TWO::equals);
  // Then
  assertEquals(5, mapped1.orElse(null));
  assertFalse(mapped2.getFailure().orElse(null));
}

Flat-Mapping Results

Flat-mapping is used to chain operations that return results themselves, flattening the nested structures into a single result object. This allows you to transform a success into a failure, or a failure into a success.

To illustrate flat-mapping concepts, the next examples will follow a familiar "pet store" theme. This involves three Java types: Pet, PetError, and PetStore. These types will help us demonstrate the effective use of flat-mapping methods.

enum PetError {NOT_FOUND, NO_CONFIG}

record Pet(long id, String name) {

  static final Pet DEFAULT = new Pet(0, "Default pet");
  static final Pet ROCKY = new Pet(1, "Rocky");
  static final Pet GARFIELD = new Pet(2, "Garfield");
}

record PetStore(Pet... pets) {

  PetStore() {
    this(Pet.ROCKY, Pet.GARFIELD);
  }

  Result<Pet, PetError> find(long id) {
    Optional<Pet> found = stream(pets).filter(pet -> pet.id() == id).findAny();
    return Results.ofOptional(found, NOT_FOUND);
  }

  Result<Pet, PetError> getDefaultPet(PetError error) {
    return error == NO_CONFIG ? success(Pet.DEFAULT) : failure(error);
  }

  Result<Long, PetError> getDefaultPetId(PetError error) {
    return getDefaultPet(error).mapSuccess(Pet::id);
  }
}

With these types defined, we'll explore how to use various flat-mapping methods to transform result objects and manage pet-related operations in our imaginary pet store.

Flat-Mapping Successful Results

Use Result::flatMapSuccess to chain an operation that returns a result object. This method applies a mapping function to the success value, replacing the original result with the new one returned by the function. If the result is a failure, it remains unchanged.

@Test
void testFlatMapSuccess() {
  // Given
  PetStore store = new PetStore();
  Result<Long, PetError> result = success(100L);
  // When
  Result<Pet, PetError> mapped = result.flatMapSuccess(store::find);
  // Then
  assertEquals(NOT_FOUND, mapped.getFailure().orElse(null));
}

This example starts with a successful result containing a wrong pet ID (not found in the pet store). When we flat-map it with the store's find method reference, the final result contains a pet error.

Flat-Mapping Failed Results

Use Result::flatMapFailure to chain a result-bearing operation. This method also replaces the original result with the new one returned by the mapping function. If the result is a success, it remains unchanged.

@Test
void testFlatMapFailure() {
  // Given
  PetStore store = new PetStore();
  Result<Long, PetError> result = failure(NO_CONFIG);
  // When
  Result<Long, PetError> mapped = result.flatMapFailure(store::getDefaultPetId);
  // Then
  assertEquals(Pet.DEFAULT.id(), mapped.orElse(null));
}

Here we start with a failed result containing a pet error. When we flat-map it with the store's getDefaultPetId method reference, the final result contains the ID of the default pet in the store.

Flat-Mapping Both Success and Failure

The Result::flatMap method handles both success and failure cases by applying the appropriate function based on the status of the original result.

@Test
void testFlatMap() {
  // Given
  PetStore store = new PetStore();
  Result<Long, PetError> result1 = success(100L);
  Result<Long, PetError> result2 = failure(NO_CONFIG);
  // When
  Result<Pet, PetError> mapped1 = result1.flatMap(store::find, store::getDefaultPet);
  Result<Pet, PetError> mapped2 = result2.flatMap(store::find, store::getDefaultPet);
  // Then
  assertEquals(NOT_FOUND, mapped1.getFailure().orElse(null));
  assertEquals(Pet.DEFAULT, mapped2.orElse(null));
}

This example starts with a successful result containing a wrong pet ID (not found in the pet store). When we flat-map it with the store's find method reference, the final result contains a pet error.

Here we start with a failed result containing a pet error. When we flat-map it with the store's getDefaultPetId method reference, the final result contains the ID of the default pet in the store.

Conclusion

We demonstrated how to transform results in a concise and functional manner, enhancing the clarity and flexibility of your error-handling and data-processing logic.

There are several ways to create result objects.

Successful Results

A successful result contains a non-null value produced by an operation when everything works as intended. We can use Results::success to create a new instance.

@Test
void testSuccess() {
  // When
  Result<Integer, ?> result = Results.success(200);
  // Then
  assertTrue(result::hasSuccess);
  assertFalse(result::hasFailure);
}

Failed Results

On the other hand, a failed result holds a value representing the problem that prevented the operation from completing. We can use Results::failure to create a new one.

@Test
void testFailure() {
  // When
  Result<?, String> result = Results.failure("The operation failed");
  // Then
  assertTrue(result::hasFailure);
  assertFalse(result::hasSuccess);
}

Results Based on Nullable Values

When we need to create results that depend on a possibly null value, we can use Results::ofNullable. If the first argument is null, then the second one will be used to create a failed result.

@Test
void testOfNullable() {
  // Given
  String string1 = "The operation succeeded";
  String string2 = null;
  // When
  Result<String, Integer> result1 = Results.ofNullable(string1, 404);
  Result<String, Integer> result2 = Results.ofNullable(string2, 404);
  // Then
  assertTrue(result1::hasSuccess);
  assertTrue(result2::hasFailure);
}

Results Based on Optionals

We can also use Results::ofOptional to create results that depend on an Optional value. If the first argument is an empty optional, then the second one will be used to create a failed result.

@Test
void testOfOptional() {
  // Given
  Optional<BigDecimal> optional1 = Optional.of(BigDecimal.ONE);
  Optional<BigDecimal> optional2 = Optional.empty();
  // When
  Result<BigDecimal, Integer> result1 = Results.ofOptional(optional1, -1);
  Result<BigDecimal, Integer> result2 = Results.ofOptional(optional2, -1);
  // Then
  assertTrue(result1::hasSuccess);
  assertTrue(result2::hasFailure);
}

Results Based on Callables

Finally, if we have a task that may either return a success value or throw an exception, we can encapsulate it as a result using Results::ofCallable so we don't need to use a try-catch block.

String task1() {
  return "OK";
}

String task2() throws Exception {
  throw new Exception("Whoops!");
}

@Test
void testOfCallable() {
  // When
  Result<String, Exception> result1 = Results.ofCallable(this::task1);
  Result<String, Exception> result2 = Results.ofCallable(this::task2);
  // Then
  assertTrue(result1::hasSuccess);
  assertTrue(result2::hasFailure);
}

Conclusion

We've covered how to create new instances of Result using various factory methods provided by the Results class. Each method serves a specific purpose, allowing you to select the most suitable one based on the situation.

This demo project demonstrates how to handle and serialize Result objects within a Spring Boot application. It provides a working example of a "pet store" web service that exposes a REST API for managing pets.

Generating the Project

The project was generated via Spring Initializr including features: web and cloud-feign.

Adding Serialization Support

Then Jackson datatype module for Result objects was manually added as a dependency to serialize and deserialize Result objects.

dependencies {
  // ...
  implementation platform('com.leakyabstractions:result-bom:1.0.0.0')
  implementation 'com.leakyabstractions:result'
  implementation 'com.leakyabstractions:result-jackson'
}

We use a @Bean to register the datatype module.

@Configuration
public class JacksonConfig {
  @Bean
  public Module registerResultModule() {
    return new ResultModule();
  }
}

API Responses

API responses contain a Result field, encapsulating the outcome of the requested operation.

public class ApiResponse<S> {

  @JsonProperty String version;
  @JsonProperty Instant generatedOn;
  @JsonProperty Result<S, ApiError> result;
}

Results have different success types, depending on the specific endpoint. Failures will be encapsulated as instances of ApiError.

Controllers

Controllers return instances of ApiResponse that will be serialized to JSON by Spring Boot.

@RestController
public class PetController {
  // ...
  @GetMapping("/pet")
  ApiResponse<Collection<Pet>> list(@RequestHeader("X-Type") RepositoryType type) {
    log.info("List all pets in {} pet store", type);
    return response(locate(type)
      .flatMapSuccess(PetRepository::listPets)
      .ifSuccess(x -> log.info("Listed {} pet(s) in {}", x.size(), type))
      .ifFailure(this::logError));
  }
}

Since failures are expressed as ApiError objects, endpoints invariably return HTTP status 200.

Running the Application

The application can be built and run with Gradle.

./gradlew bootRun

This will start a stand-alone server on port 8080.

Testing the Server

Once started, you can interact with the API.

curl -s -H 'x-type: local' http://localhost:8080/pet/0

You should see a JSON response like this:

{
  "version": "1.0",
  "result": {
    "success":{
      "id": 0,
      "name": "Rocky",
      "status": "AVAILABLE"
    }
  }
}

Using Swagger-UI

You can navigate to http://localhost:8080/ to inspect the API using an interactive UI

Lazy results optimize performance by deferring costly operations until absolutely necessary. They behave like regular results, but only execute the underlying operation when an actual check for success or failure is performed.

How to Use this Add-On

Add this Maven dependency to your build:

Creating Lazy Results

We can use LazyResults::ofSupplier to create a lazy result.

Supplier<Result<Integer, String>> supplier = () -> success(123);
Result<Integer, String> lazy = LazyResults.ofSupplier(supplier);

While suppliers can return a fixed success or failure, lazy results shine when they encapsulate time-consuming or resource-intensive operations.

/* Represents the operation we may omit */
Result<Long, Exception> expensiveCalculation(AtomicLong timesExecuted) {
  long counter = timesExecuted.incrementAndGet();
  return success(counter);
}

This sample method simply increments and returns a counter for brevity. However, in a typical scenario, this would involve an I/O operation.

Skipping Expensive Calculations

The advantage of lazy results is that they defer invoking the provided Supplier for as long as possible. Despite this, you can screen and transform them like any other result without losing their laziness.

@Test
void shouldSkipExpensiveCalculation() {
  AtomicLong timesExecuted = new AtomicLong();
  // Given
  Result<Long, Exception> lazy = LazyResults
      .ofSupplier(() -> expensiveCalculation(timesExecuted));
  // When
  Result<String, Exception> transformed = lazy.mapSuccess(Object::toString);
  // Then
  assertNotNull(transformed);
  assertEquals(0L, timesExecuted.get());
}

In this example, the expensive calculation is omitted because the lazy result is never fully evaluated. This test demonstrates that a lazy result can be transformed while maintaining laziness, ensuring that the expensive calculation is deferred.

Triggering Result Evaluation

Finally, when it's time to check whether the operation succeeds or fails, the lazy result will execute it. This is triggered by using any of the terminal methods, such as Result::hasSuccess.

@Test
void shouldExecuteExpensiveCalculation() {
  AtomicLong timesExecuted = new AtomicLong();
  // Given
  Result<Long, Exception> lazy = LazyResults
      .ofSupplier(() -> expensiveCalculation(timesExecuted));
  // When
  Result<String, Exception> transformed = lazy.mapSuccess(Object::toString);
  boolean success = transformed.hasSuccess();
  // Then
  assertTrue(success);
  assertEquals(1L, timesExecuted.get());
}

Here, the expensive calculation is executed because the lazy result is finally evaluated.

Handling Success and Failure Eagerly

By default, Result::ifSuccess, Result::ifFailure, and Result::ifSuccessOrElse are treated as terminal methods. This means they eagerly evaluate the result and then perform an action based on its status.

@Test
void shouldHandleSuccessEagerly() {
  AtomicLong timesExecuted = new AtomicLong();
  AtomicLong consumerExecuted = new AtomicLong();
  Consumer<Long> consumer = x -> consumerExecuted.incrementAndGet();
  // Given
  Result<Long, Exception> lazy = LazyResults
      .ofSupplier(() -> expensiveCalculation(timesExecuted));
  // When
  lazy.ifSuccess(consumer);
  // Then
  assertEquals(1L, timesExecuted.get());
  assertEquals(1L, consumerExecuted.get());
}

In this test, we don't explicitly unwrap the value or check the status, but since we want to consume the success value, we need to evaluate the lazy result first.

Furthermore, even if we wanted to handle the failure scenario, we would still need to evaluate the lazy result.

@Test
void shouldHandleFailureEagerly() {
  AtomicLong timesExecuted = new AtomicLong();
  AtomicLong consumerExecuted = new AtomicLong();
  Consumer<Exception> consumer = x -> consumerExecuted.incrementAndGet();
  // Given
  Result<Long, Exception> lazy = LazyResults
      .ofSupplier(() -> expensiveCalculation(timesExecuted));
  // When
  lazy.ifFailure(consumer);
  // Then
  assertEquals(1L, timesExecuted.get());
  assertEquals(0L, consumerExecuted.get());
}

In this other test, we use Result::ifFailure instead of Result::ifSuccess. Since the lazy result is evaluated to a success, the failure consumer is never executed.

Handling Success and Failure Lazily

When these conditional actions may also be skipped along with the expensive calculation, we can encapsulate them into a LazyConsumer instead of a regular Consumer. All we need to do is to create the consumer using LazyConsumer::of. Lazy consumers will preserve the laziness until a terminal method is eventually used on the result.

@Test
void shouldHandleSuccessLazily() {
  AtomicLong timesExecuted = new AtomicLong();
  AtomicLong consumerExecuted = new AtomicLong();
  Consumer<Long> consumer = LazyConsumer
      .of(x -> consumerExecuted.incrementAndGet());
  // Given
  Result<Long, Exception> lazy = LazyResults
      .ofSupplier(() -> expensiveCalculation(timesExecuted));
  // When
  lazy.ifSuccess(consumer);
  // Then
  assertEquals(0L, timesExecuted.get());
  assertEquals(0L, consumerExecuted.get());
}

Here, we use a lazy consumer with Result::ifSuccess so the expensive calculation is skipped because the lazy result is never fully evaluated.

Conclusion

We learned how to defer expensive calculations until absolutely necessary. By leveraging lazy results, you can optimize performance by avoiding unnecessary computations and only evaluating the operation's outcome when needed.

When using Result objects with Jackson we might run into some problems. The Jackson datatype module for Result solves them by making Jackson treat results as if they were ordinary objects.

How to Use this Add-On

Add this Maven dependency to your build:

Test Scenario

Let's start by creating a class ApiResponse containing one ordinary and one Result field.

/** Represents an API response */
public class ApiResponse {

  @JsonProperty
  String version;

  @JsonProperty
  Result<String, String> result;

  // Constructors, getters and setters omitted
}

Problem Overview

Then we will take a look at what happens when we try to serialize and deserialize ApiResponse objects.

Serialization Problem

Now, let's instantiate an ApiResponse object.

ApiResponse response = new ApiResponse();
response.setVersion("v1");
response.setResult(success("Perfect"));

And finally, let's try serializing it using an object mapper.

ObjectMapper objectMapper = new ObjectMapper();
String json = objectMapper.writeValueAsString(response);

We'll see that now we get an InvalidDefinitionException.

Java 8 optional type `java.util.Optional<java.lang.String>`
 not supported by default:
 add Module "com.fasterxml.jackson.datatype:jackson-datatype-jdk8"
 to enable handling

While this may look strange, it's the expected behavior. When Jackson examined the result object, it invoked Result::getSuccess and received an optional string value. But Jackson will not handle JDK 8 datatypes like Optional unless you register the appropriate modules.

@Test
void testSerializationProblem() {
  // Given
  ApiResponse response = new ApiResponse("v1", success("Perfect"));
  // Then
  ObjectMapper objectMapper = new ObjectMapper();
  InvalidDefinitionException error = assertThrows(InvalidDefinitionException.class,
      () -> objectMapper.writeValueAsString(response));
  assertTrue(error.getMessage().startsWith(
      "Java 8 optional type `java.util.Optional<java.lang.String>` not supported"));
}

This is Jackson's default serialization behavior. But we'd like to serialize the result field like this:

{
  "version": "v1",
  "result": {
    "failure": null,
    "success": "Perfect"
  }
}

Deserialization Problem

Now, let's reverse our previous example, this time trying to deserialize a JSON object into an ApiResponse.

String json = "{\"version\":\"v2\",\"result\":{\"success\":\"OK\"}}";
ObjectMapper objectMapper = new ObjectMapper();
objectMapper.readValue(json, ApiResponse.class);

We'll see that we get another InvalidDefinitionException. Let's inspect the stack trace.

Cannot construct instance of `com.leakyabstractions.result.api.Result`
 (no Creators, like default constructor, exist):
 abstract types either need to be mapped to concrete types,
 have custom deserializer, or contain additional type information

This behavior again makes sense. Essentially, Jackson cannot create new result objects because Result is an interface, not a concrete type.

@Test
void testDeserializationProblem() {
  // Given
  String json = "{\"version\":\"v2\",\"result\":{\"success\":\"OK\"}}";
  // Then
  ObjectMapper objectMapper = new ObjectMapper();
  InvalidDefinitionException error = assertThrows(InvalidDefinitionException.class,
      () -> objectMapper.readValue(json, ApiResponse.class));
  assertTrue(error.getMessage().startsWith(
      "Cannot construct instance of `com.leakyabstractions.result.api.Result`"));
}

Solution Implementation

What we want, is for Jackson to treat Result values as JSON objects that contain either a success or a failure value. Fortunately, there's a Jackson module that can solve this problem.

Registering the Jackson Datatype Module for Result

Once we have added Result-Jackson as a dependency, all we need to do is register ResultModule with our object mapper.

ObjectMapper objectMapper = new ObjectMapper();
objectMapper.registerModule(new ResultModule());

Alternatively, you can also make Jackson auto-discover the module.

objectMapper.findAndRegisterModules();

Regardless of the chosen registration mechanism, once the module is registered all functionality is available for all normal Jackson operations.

Serializing Results

Now, let's try and serialize our ApiResponse object again:

@Test
void serializeSuccessfulResult() throws Exception {
  // Given
  ApiResponse response = new ApiResponse("v3", success("All good"));
  // When
  ObjectMapper objectMapper = new ObjectMapper();
  objectMapper.registerModule(new ResultModule());
  String json = objectMapper.writeValueAsString(response);
  // Then
  assertTrue(json.contains("v3"));
  assertTrue(json.contains("All good"));
}

If we look at the serialized response, we'll see that this time the result field contains a null failure value and a non-null success value:

{
  "version": "v3",
  "result": {
    "failure": null,
    "success": "All good"
  }
}

Next, we can try serializing a failed result.

@Test
void serializeFailedResult() throws Exception {
  // Given
  ApiResponse response = new ApiResponse("v4", failure("Oops"));
  // When
  ObjectMapper objectMapper = new ObjectMapper();
  objectMapper.findAndRegisterModules();
  String json = objectMapper.writeValueAsString(response);
  // Then
  assertTrue(json.contains("v4"));
  assertTrue(json.contains("Oops"));
} // End

We can verify that the serialized response contains a non-null failure value and a null success value.

{
  "version": "v4",
  "result": {
    "failure": "Oops",
    "success": null
  }
}

Deserializing Results

Now, let's repeat our tests for deserialization. If we read our ApiResponse again, we'll see that we no longer get an InvalidDefinitionException.

@Test
void deserializeSuccessfulResult() throws Exception {
  // Given
  String json = "{\"version\":\"v5\",\"result\":{\"success\":\"Yay\"}}";
  // When
  ObjectMapper objectMapper = new ObjectMapper().findAndRegisterModules();
  ApiResponse response = objectMapper.readValue(json, ApiResponse.class);
  // Then
  assertEquals("v5", response.getVersion());
  assertEquals("Yay", response.getResult().orElse(null));
}

Finally, let's repeat the test again, this time with a failed result. We'll see that yet again we don't get an exception, and in fact, have a failed result.

@Test
void deserializeFailedResult() throws Exception {
  // Given
  String json = "{\"version\":\"v6\",\"result\":{\"failure\":\"Nay\"}}";
  // When
  ObjectMapper objectMapper = new ObjectMapper().findAndRegisterModules();
  ApiResponse response = objectMapper.readValue(json, ApiResponse.class);
  // Then
  assertEquals("v6", response.getVersion());
  assertEquals("Nay", response.getResult().getFailure().orElse(null));
}

Conclusion

We learned how to serialize and deserialize Result objects using Jackson, demonstrating how the provided datatype module enables Jackson to treat Results as ordinary objects.

When using Result objects with Micronaut, we might run into some problems. The Micronaut serialization support for Result solves them by making Micronaut treat results as Serdeable (so they can be serialized and deserialized).

How to Use this Add-On

Add this Maven dependency to your build:

Test Scenario

Let's start by creating a record ApiOperation containing one ordinary and one Result field.

/** Represents an API operation */
@Serdeable
public record ApiOperation(String name, Result<String, String> result) {
}

Problem Overview

We will take a look at what happens when we try to serialize and deserialize ApiOperation objects with Micronaut.

Serialization Problem

Now, let's create a Micronaut controller that returns an instance of ApiOperation containing a successful result.

@Controller("/operations")
public class ApiController {

    @Get("/last")
    ApiOperation lastOperation() {
        return new ApiOperation("setup", Results.success("Perfect"));
    }
}

And finally, let's run the application and try the /operations/last endpoint we just created.

curl 'http://localhost:8080/operations/last'

We'll see that we get a Micronaut CodecException caused by a SerdeException.

No serializable introspection present for type Success.
 Consider adding Serdeable. Serializable annotate to type Success.
 Alternatively if you are not in control of the project's source code,
 you can use @SerdeImport(Success.class) to enable serialization of this type.

Although this may look strange, it's actually what we should expect. Even though we annotated ApiOperation as @Serdeable, Micronaut doesn't know how to serialize result objects yet, so the data structure cannot be serialized.

@Test
void testSerializationProblem(ObjectMapper objectMapper) {
  // Given
  ApiOperation op = new ApiOperation("setup", success("Perfect"));
  // Then
  SerdeException error = assertThrows(SerdeException.class,
      () -> objectMapper.writeValueAsString(op));
  assertTrue(error.getMessage().startsWith(
      "No serializable introspection present for type Success."));
}

This is Micronaut's default serialization behavior. But we'd like to serialize the result field like this:

{
  "name": "setup",
  "result": {
    "failure": null,
    "success": "Perfect"
  }
}

Deserialization Problem

Now, let's reverse our previous example, this time trying to receive an ApiOperation as the body of a POST request.

@Controller("/operations")
public class ApiController {

    @Post("/notify")
    Map<String, String> notify(@Body ApiOperation op) {
        return op.result()
                .mapSuccess(s -> Map.of("message", op.name() + " succeeded: " + s))
                .orElseMap(f -> Map.of("error", op.name() + " failed: " + f));
    }
}

We'll see that now we get an IntrospectionException. Let's inspect the stack trace.

No bean introspection available for type
 [interface com.leakyabstractions.result.api.Result].
 Ensure the class is annotated with
 io.micronaut.core.annotation.Introspected

This behavior again makes sense. Essentially, Micronaut cannot create new result objects, because Result is not annotated as @Introspected or @Serdeable.

@Test
void testDeserializationProblem(ObjectMapper objectMapper) {
  // Given
  String json = """
      {"name":"renew","result":{"success":"OK"}}""";
  // Then
  IntrospectionException error = assertThrows(IntrospectionException.class,
      () -> objectMapper.readValue(json, ApiOperation.class));
  String errorMessage = error.getMessage(); // Extract error message
  // Verify error message
  assertTrue(errorMessage.startsWith("No bean introspection available " +
      "for type [interface com.leakyabstractions.result.api.Result]."));
} // End

Solution Implementation

What we want, is for Micronaut to treat Result values as JSON objects that contain either a success or a failure value. Fortunately, there's an easy way to solve this problem.

Adding the Serde Imports to the Classpath

All we need to do now is add Result-Micronaut-Serde as a Maven dependency. Once the @SerdeImport is in the classpath, all functionality is available for all normal Micronaut operations.

Serializing Results

Now, let's try and serialize our ApiOperation object again.

@Test
void serializeSuccessfulResult(ObjectMapper objectMapper)
    throws IOException {
  // Given
  ApiOperation op = new ApiOperation("clean", success("All good"));
  // When
  String json = objectMapper.writeValueAsString(op);
  // Then
  assertEquals("""
      {"name":"clean","result":{"success":"All good"}}""", json);
}

If we look at the serialized response, we'll see that this time the result field contains a success field.

{
  "name": "clean",
  "result": {
    "failure": null,
    "success": "All good"
  }
}

Next, we can try serializing a failed result.

@Test
void serializeFailedResult(ObjectMapper objectMapper)
    throws IOException {
  // Given
  ApiOperation op = new ApiOperation("build", failure("Oops"));
  // When
  String json = objectMapper.writeValueAsString(op);
  // Then
  assertEquals("""
      {"name":"build","result":{"failure":"Oops"}}""", json);
}

We can verify that the serialized response contains a non-null failure value and a null success value:

{
  "name": "build",
  "result": {
    "failure": "Oops",
    "success": null
  }
}

Deserializing Results

Now, let's repeat our tests for deserialization. If we read our ApiOperation again, we'll see that we no longer get an IntrospectionException.

@Test
void deserializeSuccessfulResult(ObjectMapper objectMapper)
    throws IOException {
  // Given
  String json = """
      {"name":"check","result":{"success":"Yay"}}""";
  // When
  ApiOperation response = objectMapper.readValue(json, ApiOperation.class);
  // Then
  assertEquals("check", response.name());
  assertEquals("Yay", response.result().orElse(null));
}

Finally, let's repeat the test again, this time with a failed result. We'll see that yet again we don't get an exception, and in fact, have a failed result.

@Test
void deserializeFailedResult(ObjectMapper objectMapper)
    throws IOException {
  // Given
  String json = """
      {"name":"start","result":{"failure":"Nay"}}""";
  // When
  ApiOperation response = objectMapper.readValue(json, ApiOperation.class);
  // Then
  assertEquals("start", response.name());
  assertEquals("Nay", response.result().getFailure().orElse(null));
}

Conclusion

We learned how to serialize and deserialize Result objects using Micronaut, demonstrating how the provided @SerdeImport enables Micronaut to treat Results as Serdeable objects.

You can use fluent assertions for Result objects to enhance the readability and expressiveness of your unit tests. These assertions are based on AssertJ, an open-source Java library that offers a fluent API for writing assertions in test cases.

How to Use this Add-On

Add this Maven dependency to your build:

Asserting Result Objects

You can use ResultAssertions::assertThat in your tests to create fluent assertions for result objects.

import static com.leakyabstractions.result.assertj.ResultAssertions.assertThat;

@Test
void testAssertThat() {
  // Given
  final int zero = 0;
  // When
  final Result<Integer, String> result = success(zero);
  // Then
  assertThat(zero).isZero();
  assertThat(result).hasSuccess(zero);
}

If, for any reason, you cannot statically import assertThat, you can use ResultAssert::assertThatResult instead.

import static com.leakyabstractions.result.assertj.ResultAssert.assertThatResult;
import static org.assertj.core.api.Assertions.assertThat;

@Test
void testAssertThatResult() {
  // Given
  final int zero = 0;
  // When
  final Result<Integer, String> result = success(zero);
  // Then
  assertThat(zero).isZero();
  assertThatResult(result).hasSuccess(zero);
}

Conclusion

We covered how to use fluent assertions for Results. This approach allows you to write clear and expressive tests, enhancing the maintainability of your unit tests while ensuring that Result objects behave as expected.

We'll now delve into a set of methods that allow you to take conditional actions based on the state of a result. They provide a cleaner and more expressive way to handle success and failure scenarios, eliminating the need for lengthy if/else blocks.

Handling Success

We can use Result::ifSuccess to specify an action that must be executed if the result represents a successful outcome. This method takes a consumer function that will be applied to the success value wrapped by the result.

@Test
void testIfSuccess() {
  // Given
  List<Object> list = new ArrayList<>();
  Result<Integer, String> result = success(100);
  // When
  result.ifSuccess(list::add);
  // Then
  assertEquals(100, list.getFirst());
}

In this example, ifSuccess ensures that the provided action (adding the success value to the list) is only executed if the parsing operation is successful.

Handling Failure

On the other hand, we can use Result::ifFailure method to define an action that must be taken when the result represents a failure. This method also takes a Consumer that will be applied to the failure value inside the result.

@Test
void testIfFailure() {
  // Given
  List<Object> list = new ArrayList<>();
  Result<Integer, String> result = failure("ERROR");
  // When
  result.ifFailure(list::add);
  // Then
  assertEquals("ERROR", list.getFirst());
}

Here, ifFailure ensures that the provided action (adding the failure value to the list) is only executed if the parsing operation fails.

Handling Both Scenarios

Finally, Result::ifSuccessOrElse allows you to specify two separate actions: one for when the operation succeeded and another for when it failed. This method takes two consumer functions: the first for handling the success case and the second for handling the failure case.

@Test
void testIfSuccessOrElse() {
  // Given
  List<Object> list1 = new ArrayList<>();
  List<Object> list2 = new ArrayList<>();
  Result<Long, String> result1 = success(100L);
  Result<Long, String> result2 = failure("ERROR");
  // When
  result1.ifSuccessOrElse(list1::add, list1::add);
  result2.ifSuccessOrElse(list2::add, list2::add);
  // Then
  assertEquals(100L, list1.getFirst());
  assertEquals("ERROR", list2.getFirst());
}

In this example, ifSuccessOrElse simplifies conditional logic by providing a single method to handle both success and failure scenarios, making the code more concise and readable.

Conclusion

We explained how to handle success and failure scenarios using these three methods. They provide a powerful way to perform conditional actions based on the state of a Result, streamlining your error handling and making your code more readable and maintainable.