FluentResults is a lightweight .NET library developed to solve a common problem. It returns an object indicating success or failure of an operation instead of throwing/using exceptions.
You can install FluentResults with NuGet:
Install-Package FluentResults
- Generalised container which works in all contexts (ASP.NET MVC/WebApi, WPF, DDD Domain Model, etc)
- Store multiple errors in one Result object
- Store powerful and elaborative Error and Success objects instead of only error messages in string format
- Designing Errors/Success in an object-oriented way
- Store the root cause with chain of errors in a hierarchical way
- Provide
- .NET Standard, .NET Core, .NET 5+ and .NET Full Framework support (details see .NET Targeting)
- SourceLink support
- powerful code samples which show the integration with famous or common frameworks/libraries
- FluentAssertions Extension to assert FluentResult objects in an elegant way
To be honest, the pattern - returning a Result object indicating success or failure - is not at all a new idea. This pattern comes from functional programming languages. With FluentResults this pattern can also be applied in .NET/C#.
The article Exceptions for Flow Control by Vladimir Khorikov describes very good in which scenarios the Result pattern makes sense and in which not. See the list of Best Practices and the list of resources to learn more about the Result Pattern.
A Result can store multiple Error and Success messages.
// create a result which indicates success
Result successResult1 = Result.Ok();
// create a result which indicates failure
Result errorResult1 = Result.Fail("My error message");
Result errorResult2 = Result.Fail(new Error("My error message"));
Result errorResult3 = Result.Fail(new StartDateIsAfterEndDateError(startDate, endDate));
The class Result
is typically used by void methods which have no return value.
public Result DoTask()
{
if (this.State == TaskState.Done)
return Result.Fail("Task is in the wrong state.");
// rest of the logic
return Result.Ok();
}
Additionally a value from a specific type can also be stored if necessary.
// create a result which indicates success
Result<int> successResult1 = Result.Ok(42);
Result<MyCustomObject> successResult2 = Result.Ok(new MyCustomObject());
// create a result which indicates failure
Result<int> errorResult = Result.Fail<int>("My error message");
The class Result<T>
is typically used by methods with a return type.
public Result<Task> GetTask()
{
if (this.State == TaskState.Deleted)
return Result.Fail<Task>("Deleted Tasks can not be displayed.");
// rest of the logic
return Result.Ok(task);
}
After you get a Result object from a method you have to process it. This means, you have to check if the operation was completed successfully or not. The properties IsSuccess
and IsFailed
in the Result object indicate success or failure. The value of a Result<T>
can be accessed via the properties Value
and ValueOrDefault
.
Result<int> result = DoSomething();
// get all reasons why result object indicates success or failure.
// contains Error and Success messages
IEnumerable<IReason> reasons = result.Reasons;
// get all Error messages
IEnumerable<IError> errors = result.Errors;
// get all Success messages
IEnumerable<ISuccess> successes = result.Successes;
if (result.IsFailed)
{
// handle error case
var value1 = result.Value; // throws exception because result is in failed state
var value2 = result.ValueOrDefault; // return default value (=0) because result is in failed state
return;
}
// handle success case
var value3 = result.Value; // return value and doesn't throw exception because result is in success state
var value4 = result.ValueOrDefault; // return value because result is in success state
There are many Result Libraries which store only simple string messages. FluentResults instead stores powerful object-oriented Error and Success objects. The advantage is all the relevant information of an error or success is encapsulated within one class.
The entire public api of this library uses the interfaces IReason
, IError
and ISuccess
for representing a reason, error or success. IError
and ISuccess
inherit from IReason
. If at least one IError
object exists in the Reasons
property then the result indicates a failure and the property IsSuccess
is false.
You can create your own Success
or Error
classes when you inherit from ISuccess
or IError
or if you inherit from Success
or Error
.
public class StartDateIsAfterEndDateError : Error
{
public StartDateIsAfterEndDateError(DateTime startDate, DateTime endDate)
: base($"The start date {startDate} is after the end date {endDate}")
{
Metadata.Add("ErrorCode", "12");
}
}
With this mechanism you can also create a class Warning
. You can choose if a Warning in your system indicates a success or a failure by inheriting from Success
or Error
classes.
In some cases it is necessary to chain multiple error and success messages in one result object.
var result = Result.Fail("error message 1")
.WithError("error message 2")
.WithError("error message 3")
.WithSuccess("success message 1");
Very often you have to create a fail or success result depending on a condition. Usually you can write it in this way:
var result = string.IsNullOrEmpty(firstName) ? Result.Fail("First Name is empty") : Result.Ok();
With the methods FailIf()
and OkIf()
you can also write in a more readable way:
var result = Result.FailIf(string.IsNullOrEmpty(firstName), "First Name is empty");
In some scenarios you want to execute an action. If this action throws an exception then the exception should be catched and transformed to a result object.
var result = Result.Try(() => DoSomethingCritical());
In the above example the default catchHandler is used. The behavior of the default catchHandler can be overwritten via the global Result settings (see next example). You can control how the Error object looks.
Result.Setup(cfg =>
{
cfg.DefaultTryCatchHandler = exception =>
{
if (exception is SqlTypeException sqlException)
return new ExceptionalError("Sql Fehler", sqlException);
if (exception is DomainException domainException)
return new Error("Domain Fehler")
.CausedBy(new ExceptionError(domainException.Message, domainException));
return new Error(exception.Message);
};
});
var result = Result.Try(() => DoSomethingCritical());
It is also possible to pass a custom catchHandler via the Try(..)
method.
var result = Result.Try(() => DoSomethingCritical(), ex => new MyCustomExceptionError(ex));
You can also store the root cause of the error in the error object. With the method CausedBy(...)
the root cause can be passed as Error, list of Errors, string, list of strings or as exception. The root cause is stored in the Reasons
property of the error object.
Example 1 - root cause is an exception
try
{
//export csv file
}
catch(CsvExportException ex)
{
return Result.Fail(new Error("CSV Export not executed successfully").CausedBy(ex));
}
Example 2 - root cause is an error
Error rootCauseError = new Error("This is the root cause of the error");
Result result = Result.Fail(new Error("Do something failed", rootCauseError));
Example 3 - reading root cause from errors
Result result = ....;
if (result.IsSuccess)
return;
foreach(IError error in result.Errors)
{
foreach(ExceptionalError causedByExceptionalError in error.Reasons.OfType<ExceptionalError>())
{
Console.WriteLine(causedByExceptionalError.Exception);
}
}
It is possible to add metadata to Error or Success objects.
One way of doing that is to call the method WithMetadata(...)
directly where result object is being created.
var result1 = Result.Fail(new Error("Error 1").WithMetadata("metadata name", "metadata value"));
var result2 = Result.Ok()
.WithSuccess(new Success("Success 1")
.WithMetadata("metadata name", "metadata value"));
Another way is to call WithMetadata(...)
in constructor of the Error
or Success
class.
public class DomainError : Error
{
public DomainError(string message)
: base(message)
{
WithMetadata("ErrorCode", "12");
}
}
Multiple results can be merged with the static method Merge()
.
var result1 = Result.Ok();
var result2 = Result.Fail("first error");
var result3 = Result.Ok<int>();
var mergedResult = Result.Merge(result1, result2, result3);
A result object can be converted to another result object with methods ToResult()
and ToResult<TValue>()
.
// converting a result to a result from type Result<int> with default value of int
Result.Ok().ToResult<int>();
// converting a result to a result from type Result<int> with a custom value
Result.Ok().ToResult<int>(5);
// converting a failed result to a result from type Result<int> without passing a custom value
// because result is in failed state and therefore no value is needed
Result.Fail("Failed").ToResult<int>();
// converting a result to a result from type Result<float>
Result.Ok<int>(5).ToResult<float>(v => v);
// converting a result from type Result<int> to result from type Result<float> without passing the converting
// logic because result is in failed state and therefore no converting logic needed
Result.Fail<int>("Failed").ToResult<float>();
// converting a result to a result from type Result
Result.Ok<int>().ToResult();
Within the FluentResults library in some scenarios an ISuccess, IError or IExceptionalError object is created. For example if the method Result.Fail("My Error")
is called then internally an IError object is created. If you need to overwrite this behavior and create in this scenario a custom error class then you can set the error factory via the settings. The same extension points are also available for ISuccess and IExceptionalError.
Result.Setup(cfg =>
{
cfg.SuccessFactory = successMessage => new Success(successMessage).WithMetadata("Timestamp", DateTime.Now);
cfg.ErrorFactory = errorMessage => new Error(errorMessage).WithMetadata("Timestamp", DateTime.Now);
cfg.ExceptionalErrorFactory = (errorMessage, exception) => new ExceptionalError(errorMessage ?? exception.Message, exception)
.WithMetadata("Timestamp", DateTime.Now);
});
Similar to the catch block for exceptions, the checking and handling of errors within Result object is also supported using some methods:
// check if the Result object contains an error from a specific type
result.HasError<MyCustomError>();
// check if the Result object contains an error from a specific type and with a specific condition
result.HasError<MyCustomError>(myCustomError => myCustomError.MyField == 2);
// check if the Result object contains an error with a specific metadata key
result.HasError(error => error.HasMetadataKey("MyKey"));
// check if the Result object contains an error with a specific metadata
result.HasError(error => error.HasMetadata("MyKey", metadataValue => (string)metadataValue == "MyValue"));
Checking if a result object contains a specific success object can be done with the method HasSuccess()
// check if the Result object contains a success from a specific type
result.HasSuccess<MyCustomSuccess>();
// check if the Result object contains a success from a specific type and with a specific condition
result.HasSuccess<MyCustomSuccess>(success => success.MyField == 3);
Same has the handling of errors, the checking and handling of exceptions within Result object is also supported using some methods:
// check if the Result object contains an exception from a specific type
result.HasException<MyCustomException>();
// check if the Result object contains an exception from a specific type and with a specific condition
result.HasException<MyCustomException>(MyCustomException => MyCustomException.MyField == 1);
Sometimes it is necessary to log results. First create a logger:
public class MyConsoleLogger : IResultLogger
{
public void Log(string context, string content, ResultBase result)
{
Console.WriteLine("Result: {0} {1} <{2}>", result.Reasons.Select(reason => reason.Message), content, context);
}
public void Log<TContext>(string content, ResultBase result)
{
Console.WriteLine("Result: {0} {1} <{2}>", result.Reasons.Select(reason => reason.Message), content, typeof(TContext).FullName);
}
}
Then you must register your logger in the Result settings:
var myLogger = new MyConsoleLogger();
Result.Setup(cfg => {
cfg.Logger = myLogger;
});
Finally the logger can be used on any result:
var result = Result.Fail("Operation failed")
.Log();
Additionally, a context can be passed in form of a string or of a generic type parameter. A custom message that provide more information can also be passed as content.
var result = Result.Fail("Operation failed")
.Log("logger context", "More info about the result");
var result2 = Result.Fail("Operation failed")
.Log<MyLoggerContext>("More info about the result");
You can also log results only on successes or failures:
Result<int> result = DoSomething();
result.LogIfSuccess();
result.LogIfFailed();
Try it with the power of FluentAssertions and FluentResults.Extensions.FluentAssertions
FluentResults 3.x and above supports .NET Standard 2.0 and .NET Standard 2.1. If you need support for .NET Standard 1.1, .NET 4.6.1 or .NET 4.5 use FluentResults 2.x.
Here are some samples and best practices to be followed while using FluentResult or the Result pattern in general with some famous or commonly used frameworks and libraries.
- Domain model with a command handler
- Protecting domain invariants by using for example factory methods returning a Result object
- Make each error unique by making your own custom Error classes inheriting from IError interface or Error class
- If the method doesn't have a failure scenario then don't use the Result class as return type
- Be aware that you can merge multiple failed results or return the first failed result asap
Serializing Result objects (ASP.NET WebApi, Hangfire)
- Asp.net WebController
- Hangfire Job
- Don't serialize FluentResult result objects.
- Make your own custom ResultDto class for your public api in your system boundaries
- So you can control which data is submitted and which data is serialized
- Your public api is independent of third party libraries like FluentResults
- You can keep your public api stable
MediatR request handlers returning Result objects
- Full functional .NET Core sample code with commands/queries and a ValidationPipelineBehavior
- Returns business validation errors via a Result object from a MediatR request handler back to the consumer
- Don't throw exceptions based on business validation errors
- Inject command and query validation via MediatR PipelineBehavior and return a Result object instead of throwing an exception
- Error Handling โ Returning Results by Michael Altmann
- Operation Result Pattern by Carl-Hugo Marcotte
- Exceptions for flow control in C# by Vladimir Khorikov
- Error handling: Exception or Result? by Vladimir Khorikov
- What is an exceptional situation in code? by Vladimir Khorikov
- Advanced error handling techniques by Vladimir Khorikov
- A Simple Guide by Isaac Cummings
- Flexible Error Handling w/ the Result Class by Khalil Stemmler
- Combining ASP.NET Core validation attributes with Value Objects by Vladimir Khorikov
Thanks to all the contributers and to all the people who gave feedback!
Copyright (c) Michael Altmann. See LICENSE for details.