EventFlow

Think EventFlow is great,
buy me a cup of coffee

EventFlow is a basic CQRS+ES framework designed to be easy to use.

Have a look at our getting started guide, the do’s and don’ts and the FAQ.

Features

CQRS+ES framework
Async/await first: Every part of EventFlow is written using async/await.
Highly configurable and extendable
Easy to use
No use of threads or background workers making it "web friendly"
Cancellation: All methods that does IO work or might delay execution (due to retries), takes a CancellationToken argument to allow you to cancel the operation

Examples

Complete: Shows a complete example on how to use EventFlow with in-memory event store and read models in a relatively few lines of code
Shipping: To get a more complete example of how EventFlow could be used, have a look at the shipping example found here in the code base. The example is based on the shipping example from the book "Domain-Driven Design - Tackling Complexity in the Heart of Software" by Eric Evans. Its in-progress, but should provide inspiration on how to use EventFlow on a larger scale. If you have ideas and/or comments, create a pull request or an issue

External Examples

Racetimes: Shows some features of EventFlow that are not covered in the complete example. It features entities, a read model for an entity, delete on read models, specifications and snapshots.

Overview

Here is a list of the EventFlow concepts. Use the links to navigate to the documentation.

Aggregates: Domains object that guarantees the consistency of changes being made within each aggregate
Command bus and commands: Entry point for all command/operation execution.
Event store: Storage of the event stream for aggregates. Currently there is support for these storage types.
- In-memory - only for test
- Files - only for test
- Microsoft SQL Server
- Entity Framework Core
- SQLite
- PostgreSQL
- EventStore - home page
Subscribers: Listeners that act on specific domain events. Useful if an specific action needs to be triggered after a domain event has been committed.
Read models: Denormalized representation of aggregate events optimized for reading fast. Currently there is support for these read model storage types.
- Elasticsearch
- In-memory - only for test
- Microsoft SQL Server
- Entity Framework Core
- SQLite
- PostgreSQL
Snapshots: Instead of reading the entire event stream every single time, a snapshot can be created every so often that contains the aggregate state. EventFlow supports upgrading existing snapshots, which is useful for long-lived aggregates. Snapshots in EventFlow are opt-in and EventFlow has support for
- In-memory - only for test
- Microsoft SQL Server
- Entity Framework Core
- SQLite
- PostgreSQL
Sagas: Also known as process managers, coordinates and routes messages between bounded contexts and aggregates
Queries: Value objects that represent a query without specifying how its executed, that is let to a query handler
Jobs: Perform scheduled tasks at a later time, e.g. publish a command. EventFlow provides support for these job schedulers
- Hangfire - home page
Event upgrade: As events committed to the event store is never changed, EventFlow uses the concept of event upgraders to deprecate events and replace them with new during aggregate load.
Event publishing: Sometimes you want other applications or services to consume and act on domains. For this EventFlow supports event publishing.
RabbitMQ
Metadata: Additional information for each aggregate event, e.g. the IP of the user behind the event being emitted. EventFlow ships with several providers ready to use used.
Value objects: Data containing classes used to validate and hold domain data, e.g. a username or e-mail.
Customize: Almost every single part of EventFlow can be swapped with a custom implementation through the embedded IoC container.

Complete example

Here's a complete example on how to use the default in-memory event store along with an in-memory read model.

The example consists of the following classes, each shown below

ExampleAggregate: The aggregate root
ExampleId: Value object representing the identity of the aggregate root
ExampleEvent: Event emitted by the aggregate root
ExampleCommand: Value object defining a command that can be published to the aggregate root
ExampleCommandHandler: Command handler which EventFlow resolves using its IoC container and defines how the command specific is applied to the aggregate root
ExampleReadModel: In-memory read model providing easy access to the current state

Note: This example is part of the EventFlow test suite, so checkout the code and give it a go.

[Test]
public async Task Example()
{
  // We wire up EventFlow with all of our classes. Instead of adding events,
  // commands, etc. explicitly, we could have used the the simpler
  // AddDefaults(Assembly) instead.
  using (var resolver = EventFlowOptions.New
    .AddEvents(typeof(ExampleEvent))
    .AddCommands(typeof(ExampleCommand))
    .AddCommandHandlers(typeof(ExampleCommandHandler))
    .UseInMemoryReadStoreFor<ExampleReadModel>()
    .CreateResolver())
  {
    // Create a new identity for our aggregate root
    var exampleId = ExampleId.New;

    // Resolve the command bus and use it to publish a command
    var commandBus = resolver.Resolve<ICommandBus>();
    await commandBus.PublishAsync(
      new ExampleCommand(exampleId, 42), CancellationToken.None)
      .ConfigureAwait(false);

    // Resolve the query handler and use the built-in query for fetching
    // read models by identity to get our read model representing the
    // state of our aggregate root
    var queryProcessor = resolver.Resolve<IQueryProcessor>();
    var exampleReadModel = await queryProcessor.ProcessAsync(
      new ReadModelByIdQuery<ExampleReadModel>(exampleId), CancellationToken.None)
      .ConfigureAwait(false);

    // Verify that the read model has the expected magic number
    exampleReadModel.MagicNumber.Should().Be(42);
  }
}

// The aggregate root
public class ExampleAggregate : AggregateRoot<ExampleAggregate, ExampleId>,
  IEmit<ExampleEvent>
{
  private int? _magicNumber;

  public ExampleAggregate(ExampleId id) : base(id) { }

  // Method invoked by our command
  public void SetMagicNumer(int magicNumber)
  {
    if (_magicNumber.HasValue)
      throw DomainError.With("Magic number already set");

    Emit(new ExampleEvent(magicNumber));
  }

  // We apply the event as part of the event sourcing system. EventFlow
  // provides several different methods for doing this, e.g. state objects,
  // the Apply method is merely the simplest
  public void Apply(ExampleEvent aggregateEvent)
  {
    _magicNumber = aggregateEvent.MagicNumber;
  }
}

// Represents the aggregate identity (ID)
public class ExampleId : Identity<ExampleId>
{
  public ExampleId(string value) : base(value) { }
}

// A basic event containing some information
public class ExampleEvent : AggregateEvent<ExampleAggregate, ExampleId>
{
  public ExampleEvent(int magicNumber)
  {
      MagicNumber = magicNumber;
  }

  public int MagicNumber { get; }
}

// Command for update magic number
public class ExampleCommand : Command<ExampleAggregate, ExampleId>
{
  public ExampleCommand(
    ExampleId aggregateId,
    int magicNumber)
    : base(aggregateId)
  {
    MagicNumber = magicNumber;
  }

  public int MagicNumber { get; }
}

// Command handler for our command
public class ExampleCommandHandler
  : CommandHandler<ExampleAggregate, ExampleId, ExampleCommand>
{
  public override Task ExecuteAsync(
    ExampleAggregate aggregate,
    ExampleCommand command,
    CancellationToken cancellationToken)
  {
    aggregate.SetMagicNumer(command.MagicNumber);
    return Task.FromResult(0);
  }
}

// Read model for our aggregate
public class ExampleReadModel : IReadModel,
  IAmReadModelFor<ExampleAggregate, ExampleId, ExampleEvent>
{
  public int MagicNumber { get; private set; }

  public void Apply(
    IReadModelContext context,
    IDomainEvent<ExampleAggregate, ExampleId, ExampleEvent> domainEvent)
  {
    MagicNumber = domainEvent.AggregateEvent.MagicNumber;
  }
}

State of EventFlow

EventFlow is still under development, especially the parts regarding how read models are re-populated.

EventFlow is currently used in production environments and performs very well, but it need to mature before key APIs are stable.

EventFlow is greatly opinionated, but its possible to create new implementations for almost every part of EventFlow by registering a different implementation of an interface.

Useful articles related to EventFlow and DDD

Many of the technical design decisions in EventFlow is based on articles. This section lists some of them. If you have a link with a relevant article, please share it by creating an issue with the link.

Domain-Driven Design
General CQRS+ES
- CQRS Journey by Microsoft published by Microsoft
- An In-Depth Look At CQRS by Mike Mogosanu
- CQRS, Task Based UIs, Event Sourcing agh! by Greg Young
- Busting some CQRS myths by Jimmy Bogard
- CQRS applied by Gabriel Schenker
- DDD Decoded - Entities and Value Objects Explained
Eventual consistency
- How To Ensure Idempotency In An Eventual Consistent DDD/CQRS Application by Mike Mogosanu
- DDD Decoded - Don't Fear Eventual Consistency
Why not to implement "unit of work" in DDD
- Unit Of Work is the new Singleton by Mike Mogosanu
- The Unit of Work and Transactions In Domain Driven Design by Mike Mogosanu

Integration tests

EventFlow has several tests that verify that its able to use the systems it integrates with correctly.

Elasticsearch: Elasticsearch run as Docker Windows Container. if use in local, requires its environment and docker-compose tool, and execute PS> up_integration-test-env.ps1
EventStore: EventStore is same as the above
RabbitMQ: RabbitMQ is same as the above
MSSQL: Microsoft SQL Server is required to be running
RabbitMQ: Set an environment variable named RABBITMQ_URL with the URL for the RabbitMQ instance you would like to use.
EntityFramework: Microsoft SQL Server and PostgreSQL is required to be running
PostgreSQL: PostgreSQL is required to be running

There's a Vagrant box with both Elasticsearch and RabbitMQ you can use here.

Alternatively you can skip the NUnit tests marked with the integration category.

Thanks

Contributors
JetBrains: OSS licenses
olholm: Current logo
iconmonstr: First logo
JC008: License for Navicat Essentials for SQLite

License

The MIT License (MIT)

Copyright (c) 2015-2018 Rasmus Mikkelsen
Copyright (c) 2015-2018 eBay Software Foundation
https://github.com/eventflow/EventFlow

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

AnthonyCarl/EventFlow