Earlier this year Mike Wasson has published a post: "Dependency Injection in ASP.NET Web API 2" that describes Web API's approach to the Dependency Injection design pattern.

In short it goes like this:

• Web API provides a primary integration point through HttpConfiguration.DependencyResolver property, and tries to obtain many services through this resolver;
• Web API suggests to use your favorite Dependecy Injection library through the integration point. Author lists following libraries: Unity (by Microsoft), Castle Windsor, Spring.Net, Autofac, Ninject, and StructureMap.

The Unity Container (Unity) is a lightweight, extensible dependency injection container. There are Nugets both for Unity library and for Web API integration.

Now to the point of this post.

Unity defines a hierarchy of injection scopes. In Web API they are usually mapped to application and request scopes. This way a developer can inject application singletons, create request level, or transient objects.

Everything looks reasonable. The only problem we have found is that there is no way you to inject Web API objects like HttpConfiguration, HttpControllerContext or request's CancellationToken, as they are never registered for injection.

To workaround this we have created a small class called UnityControllerActivator that perfroms required registration:

using System;
using System.Net.Http;
using System.Threading;
using System.Threading.Tasks;
using System.Web.Http.Controllers;
using System.Web.Http.Dispatcher;

using Microsoft.Practices.Unity;

/// <summary>
/// Unity controller activator.
/// </summary>
public class UnityControllerActivator: IHttpControllerActivator
{
  /// <summary>
  /// Creates an UnityControllerActivator instance.
  /// </summary>
  /// <param name="activator">Base activator.</param>
  public UnityControllerActivator(IHttpControllerActivator activator)
  {
    if (activator == null)
    {
      throw new ArgumentException("activator");
    }

    this.activator = activator;
  }

  /// <summary>
  /// Creates a controller wrapper.
  /// </summary>
  /// <param name="request">A http request.</param>
  /// <param name="controllerDescriptor">Controller descriptor.</param>
  /// <param name="controllerType">Controller type.</param>
  /// <returns>A controller wrapper.</returns>
  public IHttpController Create(
    HttpRequestMessage request,
    HttpControllerDescriptor controllerDescriptor,
    Type controllerType)
  {
    return new Controller
    {
      activator = activator,
      controllerType = controllerType
    };
  }

  /// <summary>
  /// Base controller activator.
  /// </summary>
  private readonly IHttpControllerActivator activator;

  /// <summary>
  /// A controller wrapper.
  /// </summary>
  private class Controller: IHttpController, IDisposable
  {
    /// <summary>
    /// Base controller activator.
    /// </summary>
    public IHttpControllerActivator activator;

    /// <summary>
    /// Controller type.
    /// </summary>
    public Type controllerType;

    /// <summary>
    /// A controller instance.
    /// </summary>
    public IHttpController controller;

    /// <summary>
    /// Disposes controller.
    /// </summary>
    public void Dispose()
    {
      var disposable = controller as IDisposable;

      if (disposable != null)
      {
        disposable.Dispose();
      }
    }

    /// <summary>
    /// Executes an action.
    /// </summary>
    /// <param name="controllerContext">Controller context.</param>
    /// <param name="cancellationToken">Cancellation token.</param>
    /// <returns>Response message.</returns>
    public Task<HttpResponseMessage> ExecuteAsync(
      HttpControllerContext controllerContext,
      CancellationToken cancellationToken)
    {
      if (controller == null)
      {
        var request = controllerContext.Request;
        var container = request.GetDependencyScope().
          GetService(typeof(IUnityContainer)) as IUnityContainer;

        if (container != null)
        {
          container.RegisterInstance<HttpControllerContext>(controllerContext);
          container.RegisterInstance<HttpRequestMessage>(request);
          container.RegisterInstance<CancellationToken>(cancellationToken);
        }

        controller = activator.Create(
          request,
          controllerContext.ControllerDescriptor,
          controllerType);
      }

      controllerContext.Controller = controller;

      return controller.ExecuteAsync(controllerContext, cancellationToken);
    }
  }
}

Note on how it works.

• IHttpControllerActivator is a controller factory, which Web API uses to create new controller instances using IHttpControllerActivator.Create(). Later controller's IHttpController.ExecuteAsync() is called to run the logic.
• UnityControllerActivator replaces original controller activator with a wrapper that delays creation (injection) of real controller untill request objects are registered in the scope

To register this class one need to update code in the UnityWebApiActivator.cs (file added with nuget Unity.AspNet.WebApi)

public static class UnityWebApiActivator
{
  /// <summary>Integrates Unity when the application starts.<summary>
  public static void Start()
  {
    var config = GlobalConfiguration.Configuration;
    var container = UnityConfig.GetConfiguredContainer();

    container.RegisterInstance<HttpConfiguration>(config);
    container.RegisterInstance<IHttpControllerActivator>(
      new UnityControllerActivator(config.Services.GetHttpControllerActivator()));

    config.DependencyResolver = UnityHierarchicalDependencyResolver(container);
  }
  ...
}

With this addition we have simplified the boring problem with passing of CancellationToken all around the code, as controller (and other classes) just declared a property to inject:

public class MyController: ApiController
{
  [Dependency]
  public CancellationToken CancellationToken { get; set; }

  [Dependency]
  public IModelContext Model { get; set; }

  public async Task<IEnumerable<Products>> GetProducts(...)
  {
    ...
  }

  public async Task<IEnumerable<Customer>> GetCustomer(...)
  {
    ...
  }

  ...
}

...

public class ModelContext: IModelContext
{
  [Dependency]
  public CancellationToken CancellationToken { get; set; }
  ...
}

And finally to perform unit tests for controllers with Depenency Injection you can use a code like this:

using System.Threading;
using System.Threading.Tasks;
using System.Web.Http;
using System.Web.Http.Controllers;
using System.Web.Http.Dependencies;
using System.Net.Http;

using Microsoft.Practices.Unity;
using Microsoft.Practices.Unity.WebApi;

using Microsoft.VisualStudio.TestTools.UnitTesting;

[TestClass]
public class MyControllerTest
{
  [ClassInitialize]
  public static void Initialize(TestContext context)
  {
    config = new HttpConfiguration();

    Register(config);
  }

  [ClassCleanup]
  public static void Cleanup()
  {
    config.Dispose();
  }

  [TestMethod]
  public async Task GetProducts()
  {
    var controller = CreateController<MyController>();

    //...
  }

  public static T CreateController<T>(HttpRequestMessage request = null)
    where T: ApiController
  {
    if (request == null)
    {
      request = new HttpRequestMessage();
    }

    request.SetConfiguration(config);

    var controllerContext = new HttpControllerContext()
    {
      Configuration = config,
      Request = request
    };

    var scope = request.GetDependencyScope();
    var container = scope.GetService(typeof(IUnityContainer))
      as IUnityContainer;

    if (container != null)
    {
      container.RegisterInstance<HttpControllerContext>(controllerContext);
      container.RegisterInstance<HttpRequestMessage>(request);
      container.RegisterInstance<CancellationToken>(CancellationToken.None);
    }

    T controller = scope.GetService(typeof(T)) as T;

    controller.Configuration = config;
    controller.Request = request;
    controller.ControllerContext = controllerContext;

    return controller;
  }

  public static void Register(HttpConfiguration config)
  {
    config.DependencyResolver = CreateDependencyResolver(config);
  }

  public static IDependencyResolver CreateDependencyResolver(HttpConfiguration config)
  {
    var container = new UnityContainer();

    container.RegisterInstance<HttpConfiguration>(config);

    // TODO: configure Unity contaiener.

    return new UnityHierarchicalDependencyResolver(container);
  }

  public static HttpConfiguration config;
}

P.S. To those who think Dependency Injection is an universal tool, please read the article: Dependency Injection is Evil.

Earlier this year Mike Wasson has published a post: "Dependency Injection in ASP.NET Web API 2" that describes Web API's approach to the Dependency Injection design pattern.

In short it goes like this:

• Web API provides a primary integration point through HttpConfiguration.DependencyResolver property, and tries to obtain many services through this resolver;
• Web API suggests to use your favorite Dependecy Injection library through the integration point. Author lists following libraries: Unity (by Microsoft), Castle Windsor, Spring.Net, Autofac, Ninject, and StructureMap.

The Unity Container (Unity) is a lightweight, extensible dependency injection container. There are Nugets both for Unity library and for Web API integration.

Now to the point of this post.

Unity defines a hierarchy of injection scopes. In Web API they are usually mapped to application and request scopes. This way a developer can inject application singletons, create request level, or transient objects.

Everything looks reasonable. The only problem we have found is that there is no way you to inject Web API objects like HttpConfiguration, HttpControllerContext or request's CancellationToken, as they are never registered for injection.

To workaround this we have created a small class called UnityControllerActivator that perfroms required registration:

using System;
using System.Net.Http;
using System.Threading;
using System.Threading.Tasks;
using System.Web.Http.Controllers;
using System.Web.Http.Dispatcher;

using Microsoft.Practices.Unity;

/// <summary>
/// Unity controller activator.
/// </summary>
public class UnityControllerActivator: IHttpControllerActivator
{
  /// <summary>
  /// Creates an UnityControllerActivator instance.
  /// </summary>
  /// <param name="activator">Base activator.</param>
  public UnityControllerActivator(IHttpControllerActivator activator)
  {
    if (activator == null)
    {
      throw new ArgumentException("activator");
    }

    this.activator = activator;
  }

  /// <summary>
  /// Creates a controller wrapper.
  /// </summary>
  /// <param name="request">A http request.</param>
  /// <param name="controllerDescriptor">Controller descriptor.</param>
  /// <param name="controllerType">Controller type.</param>
  /// <returns>A controller wrapper.</returns>
  public IHttpController Create(
    HttpRequestMessage request,
    HttpControllerDescriptor controllerDescriptor,
    Type controllerType)
  {
    return new Controller
    {
      activator = activator,
      controllerType = controllerType
    };
  }

  /// <summary>
  /// Base controller activator.
  /// </summary>
  private readonly IHttpControllerActivator activator;

  /// <summary>
  /// A controller wrapper.
  /// </summary>
  private class Controller: IHttpController, IDisposable
  {
    /// <summary>
    /// Base controller activator.
    /// </summary>
    public IHttpControllerActivator activator;

    /// <summary>
    /// Controller type.
    /// </summary>
    public Type controllerType;

    /// <summary>
    /// A controller instance.
    /// </summary>
    public IHttpController controller;

    /// <summary>
    /// Disposes controller.
    /// </summary>
    public void Dispose()
    {
      var disposable = controller as IDisposable;

      if (disposable != null)
      {
        disposable.Dispose();
      }
    }

    /// <summary>
    /// Executes an action.
    /// </summary>
    /// <param name="controllerContext">Controller context.</param>
    /// <param name="cancellationToken">Cancellation token.</param>
    /// <returns>Response message.</returns>
    public Task<HttpResponseMessage> ExecuteAsync(
      HttpControllerContext controllerContext,
      CancellationToken cancellationToken)
    {
      if (controller == null)
      {
        var request = controllerContext.Request;
        var container = request.GetDependencyScope().
          GetServices(typeof(IUnityContainer)) as IUnityContainer;

        if (container != null)
        {
          container.RegisterInstance<HttpControllerContext>(controllerContext);
          container.RegisterInstance<HttpRequestMessage>(request);
          container.RegisterInstance<CancellationToken>(cancellationToken);
        }

        controller = activator.Create(
          request,
          controllerContext.ControllerDescriptor,
          controllerType);
      }

      controllerContext.Controller = controller;

      return controller.ExecuteAsync(controllerContext, cancellationToken);
    }
  }
}

Note on how it works.

• IHttpControllerActivator is a controller factory, which Web API uses to create new controller instances using IHttpControllerActivator.Create(). Later controller's IHttpController.ExecuteAsync() is called to run the logic.
• UnityControllerActivator replaces original controller activator with a wrapper that delays creation (injection) of real controller untill request objects are registered in the scope

To register this class one need to update code in the UnityWebApiActivator.cs (file added with nuget Unity.AspNet.WebApi)

public static class UnityWebApiActivator
{
  /// <summary>Integrates Unity when the application starts.<summary>
  public static void Start()
  {
    var config = GlobalConfiguration.Configuration;
    var container = UnityConfig.GetConfiguredContainer();

    container.RegisterInstance<HttpConfiguration>(config);
    container.RegisterInstance<IHttpControllerActivator>(
      new UnityControllerActivator(config.Services.GetHttpControllerActivator()));

    config.DependencyResolver = UnityHierarchicalDependencyResolver(container);
  }
  ...
}

With this addition we have simplified the boring problem with passing of CancellationToken all around the code, as controller (and other classes) just declared a property to inject:

public class MyController: ApiController
{
  [Dependency]
  public CancellationToken CancellationToken { get; set; }

  [Dependency]
  public IModelContext Model { get; set; }

  public async Task<IEnumerable<Products>> GetProducts(...)
  {
    ...
  }

  public async Task<IEnumerable<Customer>> GetCustomer(...)
  {
    ...
  }

  ...
}

...

public class ModelContext: IModelContext
{
  [Dependency]
  public CancellationToken CancellationToken { get; set; }
  ...
}

And finally to perform unit tests for controllers with Depenency Injection you can use a code like this:

using System.Threading;
using System.Threading.Tasks;
using System.Web.Http;
using System.Web.Http.Controllers;
using System.Web.Http.Dependencies;
using System.Net.Http;

using Microsoft.Practices.Unity;
using Microsoft.Practices.Unity.WebApi;

using Microsoft.VisualStudio.TestTools.UnitTesting;

[TestClass]
public class MyControllerTest
{
  [ClassInitialize]
  public static void Initialize(TestContext context)
  {
    config = new HttpConfiguration();

    Register(config);
  }

  [ClassCleanup]
  public static void Cleanup()
  {
    config.Dispose();
  }

  [TestMethod]
  public async Task GetProducts()
  {
    var controller = CreateController<MyController>();

    //...
  }

  public static T CreateController<T>(HttpRequestMessage request = null)
    where T: ApiController
  {
    if (request == null)
    {
      request = new HttpRequestMessage();
    }

    request.SetConfiguration(config);

    var controllerContext = new HttpControllerContext()
    {
      Configuration = config,
      Request = request
    };

    var scope = request.GetDependencyScope();
    var container = scope.GetService(typeof(IUnityContainer))
      as IUnityContainer;

    if (container != null)
    {
      container.RegisterInstance<HttpControllerContext>(controllerContext);
      container.RegisterInstance<HttpRequestMessage>(request);
      container.RegisterInstance<CancellationToken>(CancellationToken.None);
    }

    T controller = scope.GetService(typeof(T)) as T;

    controller.Configuration = config;
    controller.Request = request;
    controller.ControllerContext = controllerContext;

    return controller;
  }

  public static void Register(HttpConfiguration config)
  {
    config.DependencyResolver = CreateDependencyResolver(config);
  }

  public static IDependencyResolver CreateDependencyResolver(HttpConfiguration config)
  {
    var container = new UnityContainer();

    container.RegisterInstance<HttpConfiguration>(config);

    // TODO: configure Unity contaiener.

    return new UnityHierarchicalDependencyResolver(container);
  }

  public static HttpConfiguration config;
}

P.S. To those who think Dependency Injection is an universal tool, please read the article: Dependency Injection is Evil.

Farewell Entity Framework and hello Dapper!

For many years we were using Entity Framework. It's still very popular and Microsoft's primary Object-Relational Mapper library.

Clearly, the decision is subjective but here are our arguments.

We know and love SQL, and think that in its domain it occupies strong positions. What SQL leaves out of scope is a bridge between itself and other languages. That's where ORM should help.

We strongly beleive that no ORM library should try to hide SQL behind the Object's language itself. We beleive in a separation of roles in development. Database design and Data Access Layer should be separated from client's logic. Thus, we strive, if possible, to encapulate data access through SQL functions and stored procedures.

Entity Framework, in contrast, tries to factor out SQL, giving a perspective of object graph to a client. Initially, it looks promising but at the end a developer should remember that any object query is mapped back to SQL. Without keeping this in mind either query won't compile, or performance will be poor.

E.g. This query will probably fail to build SQL, as no Regex can be mapped to SQL:

var result = context.Content.
  Where(data => Regex.IsMatch(data.Content, pattern)).
  ToArray();

This query might be slow, if no suitble SQL index is defined:

var result = context.Content.
  Where(data => data.Field == value).
  ToArray();

Thus no EF's goal is achieved completely, SQL power is limitted, and Data Access Layer is often fused into other client's logic.

We think that Entity Framework is over-engineered library, which tries to be more than ORM. Its generality often bumps into limits of SQL support in EF: SQL dialects, types, operators, functions, and so on. One can observe that people for years appeal to introduce support of xml, hierarchyid, geometry/geography types, full text search, and so on. This state cannot be different, as EF will never be able and does not aim to support all SQL features.

EF has both design-time and runtime. Each database vendor should implement their EF adapter for EF to play well with that database. This cooperation is not always smooth. E.g see Database first create entity framework 6.1.1 model using system.data.sqlite 1.0.93.

At some point the cost of dealing with EF has became too high for us, so we started to look into an alternatives: from plain ADO.NET to lighter ORM library.

To our delight we have immediately found: Dapper - a simple object mapper for .NET. It provides a simple extensions to IDBConnection interface to deal with mapping of query parameters to object properties, and of query results to plain types. Here are some examples:

// Get Customer
var customer = connection.
  Query<Customer>("select * from Customers where CustomerId = @id", new { id = customerID }).
  ToSingle();

// Insert a value
connection.Execute("insert into MyTable(A, B) values(@a, @b)", new { a = 2, b = 3 });

So, Dapper leaves you with plain SQL, which we consider as advantage.

Except beeing minimalistic compared to EF, Dapper claims performance close to pure hand written ADO.NET. Indeed, they build dynamic methods to populate parameters and to create rows instances, so reflection is used during warm up period only.

After investigation we have found that Saxon 9.6 HE does not support xslt 3.0 as we assumed earlier.

On Saxonica site it's written: "Support for XQuery 3.0 and XPath 3.0 (now Recommendations) has been added to the open-source product."

As one can notice no xslt is mentioned.

More details are on open-source 3.0 support.

The new release of Saxon HE (version 9.6) claims basic support of xslt 3.0. So we're eager to test it but... errors happen. See error report at Error in SaxonHE9-6-0-1J and Bug #2160.

As with previous release Exception during execution in Saxon-HE-9.5.1-6 we bumped into engine's internal error.

We expect to see an update very soon, and to continue with long waited xslt 3.0.

Here is an argument to the discussion of open source vs commercial projects: open source projects with rich community may benefit, as problems are detected promptly; while commercial projects risk to live with more unnoticed bugs.

With Saxon 9.6 we can finally play with open source xslt 3.0

It's sad that it took so much time to make it available.

See Saxonica's home page to get details.

Among latest C++ proposals the most ambiguous is N4021.

The goal of that proposal is "to define a 2D drawing API for the C++ programming language".

The motivation is going like this:

Today, computer graphics are pervasive in modern life, and are even replacing console-style I/O for basic user interaction on many platforms. For example, a simple cout << "Hello, world!" statement doesn’t do anything useful on many tablets and smartphones. We feel that C++ programmers should have a simple, standard way of displaying 2D graphics to users.

Authors compare several public and proprietary APIs to select the one named cairo graphics library as a base.

Reflecting on starting point they write:

Taken as a whole, starting from cairo allows for the creation of a 2D C++ drawing library that is already known to be portable, implementable, and useful without the need to spend years drafting, implementing, and testing a library to make sure that it meets those criteria.
...
An alternative design would be to create a new API via a synthesis of existing 2D APIs. This has the benefit of being able to avoid any perceived design flaws that existing APIs suffer from. Unfortunately this would not have implementation and usage experience. Further, doing so would not provide any guarantee that design flaws would not creep in.

What follows is a discussion on best way to transform that C library into std style C++ API.

Our thoughts on this proposal are threefold:

1. This proposal seems a decade or two late.
2. C++ standard should be modular to support basic and optional features.
3. We feel that programmers will not be satisfied with bare 2D graphics. It's not enough at nowadays.

Indeed, appeals to create standard C++ API for UI are as old as the C++'s standardization process. It's clear why did the committee not produce such API yet: they are bureaucracy that can approve API only. In fact it's a role of community to invent and implement libraries that may make their way into the standard. Without consensus in community no standard will reflect such API.

On the other hand C++ spec at present is too fat. Probably, not many people are satisfied with the pace of its evolution. Any big chunk of a new API makes the progress even slower. C++ spec should go through a refactoring and be split into core(s) and libraries and to allow individual progress of each part. This will simplify both specification and implementation. After that refactoring an API can be added or deprecated much more easily. In fact implementations were always like this. It's the spec that tries to be monolith.

As for a new 2D graphics API. It looks like an idea from late 90-es. We think that today's programmers (at least several samples ) wished to deal with industry standard UI API, and not to start from basic drawing. Looking around we observe that html 5 is such de-facto standard. Take into an account that it supports rich layout, svg, canvas, user input; in addition it's good for GPU optimization. Even if you want to deal with simple graphics then you can build svg markup or draw on the canvas.

So, what we rather prefer to see in the C++ spec is an html binding API (both for DOM and Javascript).

Just think of standard C++ program that uses html engine as its UI!

Looking at Guava Cache we think its API is more convenient than .NET's Cache API.

Just consider:

• .NET has getters, and setters of objects by string keys.
  You should provide caching policy with each setter.

• Guava cache operates with typed storage of Key to Value.
  Provides a value factory and a caching policy in advance at cache construction.

Guava's advantange is based on an idea that homogenous storage assumes a uniform way of creation of values, and uniform caching policy. Thus a great part of logic is factored out into a cache initialization.

We have decided to create a simple adapter of the MemoryCache to achieve the same goal. Here is a result of such an experiment:

public class Cache<K, V>
  where V: class
{
  /// <summary>
  /// A cache builder.
  /// </summary>
  public struct Builder
  {
    /// <summary>
    /// A memory cache. If not specified then MemoryCache.Default is used.
    /// </summary>
    public MemoryCache MemoryCache;

    /// <summary>
    /// An expiration value.
    /// Alternatively CachePolicyFunc can be used.
    /// </summary>
    public TimeSpan Expiration;

    /// <summary>
    /// Indicates whether to use sliding (true), or absolute (false)
    /// expiration.
    /// Alternatively CachePolicyFunc can be used.
    /// </summary>
    public bool Sliding;

    /// <summary>
    /// Optional function to get caching policy.
    /// Alternatively Expiration and Sliding property can be used.
    /// </summary>
    public Func<V, CacheItemPolicy> CachePolicyFunc;

    /// <summary>
    /// Optional value validator.
    /// </summary>
    public Func<V, bool> Validator;

    /// <summary>
    /// A value factory.
    /// Alternatively FactoryAsync can be used.
    /// </summary>
    public Func<K, V> Factory;

    /// <summary>
    /// Async value factory.
    /// Alternatively Factory can be used.
    /// </summary>
    public Func<K, Task<V>> FactoryAsync;

    /// <summary>
    /// A key to string converter.
    /// </summary>
    public Func<K, string> KeyFunc;

    /// <summary>
    /// Converts builder to a Cache<K, V> instance.
    /// </summary>
    /// <param name="builder">A builder to convert.</param>
    /// <returns>A Cache<K, V> instance.</returns>
    public static implicit operator Cache<K, V>(Builder builder)
    {
      return new Cache<K, V>(builder);
    }
  }

  /// <summary>
  /// Creates a cache from a cache builder.
  /// </summary>
  /// <param name="builder">A cache builder instance.</param>
  public Cache(Builder builder)
  {
    if ((builder.Factory == null) && (builder.FactoryAsync == null))
    {
      throw new ArgumentException("builder.Factory");
    }

    if (builder.MemoryCache == null)
    {
      builder.MemoryCache = MemoryCache.Default;
    }

    this.builder = builder;
  }

  /// <summary>
  /// Cached value by key.
  /// </summary>
  /// <param name="key">A key.</param>
  /// <returns>A cached value.</returns>
  public V this[K key]
  {
    get { return Get(key); }
    set { Set(key, value); }
  }

  /// <summary>
  /// Sets a value for a key.
  /// </summary>
  /// <param name="key">A key to set.</param>
  /// <param name="value">A value to set.</param>
  public void Set(K key, V value)
  {
    SetImpl(GetKey(key), IsValid(value) ? value : null);
  }

  /// <summary>
  /// Gets a value for a key.
  /// </summary>
  /// <param name="key">A key to get value for.</param>
  /// <returns>A value instance.</returns>
  public V Get(K key)
  {
    var keyValue = GetKey(key);
    var value = builder.MemoryCache.Get(keyValue) as V;

    if (!IsValid(value))
    {
      value = CreateValue(key);
      SetImpl(keyValue, value);
    }

    return value;
  }

  /// <summary>
  /// Gets a task to return an async value.
  /// </summary>
  /// <param name="key">A key.</param>
  /// <returns>A cached value.</returns>
  public async Task<V> GetAsync(K key)
  {
    var keyValue = GetKey(key);
    var value = builder.MemoryCache.Get(keyValue) as V;

    if (!IsValid(value))
    {
      value = await CreateValueAsync(key);
      SetImpl(keyValue, value);
    }

    return value;
  }

  /// <summary>
  /// Gets string key value for a key.
  /// </summary>
  /// <param name="key">A key.</param>
  /// <returns>A string key value.</returns>
  protected string GetKey(K key)
  {
    return builder.KeyFunc != null ? builder.KeyFunc(key) :
      key == null ? null : key.ToString();
  }

  /// <summary>
  /// Creates a value for a key.
  /// </summary>
  /// <param name="key">A key to create value for.</param>
  /// <returns>A value instance.</returns>
  protected V CreateValue(K key)
  {
    return builder.Factory != null ? builder.Factory(key) :
      builder.FactoryAsync(key).Result;
  }

  /// <summary>
  /// Creates a task for value for a key.
  /// </summary>
  /// <param name="key">A key to create value for.</param>
  /// <returns>A task for a value instance.</returns>
  protected Task<V> CreateValueAsync(K key)
  {
    return builder.FactoryAsync != null ? builder.FactoryAsync(key) :
      Task.FromResult(builder.Factory(key));
  }

  /// <summary>
  /// Validates the value.
  /// </summary>
  /// <param name="value">A value to validate.</param>
  /// <returns>
  /// true if value is valid for a cache, and false otherise.
  /// </returns>
  protected bool IsValid(V value)
  {
    return (value != null) &&
      ((builder.Validator == null) || builder.Validator(value));
  }

  /// <summary>
  /// Set implementation.
  /// </summary>
  /// <param name="key">A key to set value for.</param>
  /// <param name="value">A value to set.</param>
  /// <returns>A set value.</returns>
  private V SetImpl(string key, V value)
  {
    if (value == null)
    {
      builder.MemoryCache.Remove(key);
    }
    else
    {
      builder.MemoryCache.Set(
        key,
        value,
        builder.CachePolicyFunc != null ? builder.CachePolicyFunc(value) :
        builder.Sliding ?
          new CacheItemPolicy { SlidingExpiration = builder.Expiration } :
          new CacheItemPolicy
          {
            AbsoluteExpiration = DateTime.Now + builder.Expiration
          });
    }

    return value;
  }

  /// <summary>
  /// Cache builder.
  /// </summary>
  private Builder builder;
}


The use consists of initialization:

Cache<MyKey, MyValue> MyValues = new Cache<MyKey, MyValue>.Builder
{
  KeyFunc = key => ...key to string value...,
  Factory = key => ...create a value for a key...,
  Expiration = new TimeSpan(0, 3, 0),
  Sliding = true
};

and a trivial cache access:

var value = MyValues[key];

This contrasts with MemoryCache coding pattern:

MemoryCache cache = MemoryCache.Default;
...

var keyAsString = ...key to string value...
var value = cache.Get(keyAsString) as MyValue;

if (value == null)
{
  value = ...create a value for a key...

  cache.Set(keyAsString, value, ...caching policy...);
}

These days we're not active xslt developers, though we still consider xslt and xquery are important part of our personal experience and self education.

Besides, we have a pretty large xslt code base that is and will be in use. We think xslt/xquery is in use in many other big and small projects thus they have a strong position as a niche languages.

Thus we think it's important to help to those who support xslt/xquery engines. That's what we're regularly doing (thanks to our code base).

Now, to the problem we just have found. Please consider the code:

<xsl:stylesheet version="2.0"
  xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  xmlns:xs="http://www.w3.org/2001/XMLSchema"
  xmlns:t="this"
  exclude-result-prefixes="xs t">

  <xsl:template match="/">
    <xsl:param name="new-line-text" as="xs:string" select="' '"/>
    <xsl:variable name="items" as="item()*" select="'select', $new-line-text"/>

    <xsl:message select="t:string-join($items)"/>
  </xsl:template>

  <!--
    Joins adjacent string items.
      $items - items to join.
      Returns joined items.
  -->
  <xsl:function name="t:string-join" as="item()*">
    <xsl:param name="items" as="item()*"/>

    <xsl:variable name="indices" as="xs:integer*" select="
      0,
      index-of
      (
        (
          for $item in $items return
            $item instance of xs:string
        ),
        false()
      ),
      count($items) + 1"/>

    <xsl:sequence select="
      for $i in 1 to count($indices) - 1 return
      (
        $items[$indices[$i]],
        string-join
        (
          subsequence
          (
            $items,
            $indices[$i] + 1,
            $indices[$i + 1] - $indices[$i] - 1
          ),
          ''
        )
      )"/>
  </xsl:function>

</xsl:stylesheet>

The output is:

Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: -1
  at net.sf.saxon.om.Chain.itemAt(Chain.java:161)
  at net.sf.saxon.om.SequenceTool.itemAt(SequenceTool.java:130)
  at net.sf.saxon.expr.FilterExpression.iterate(FilterExpression.java:1143)
  at net.sf.saxon.expr.LetExpression.iterate(LetExpression.java:365)
  at net.sf.saxon.expr.instruct.BlockIterator.next(BlockIterator.java:49)
  at net.sf.saxon.expr.MappingIterator.next(MappingIterator.java:70)
  at net.sf.saxon.expr.instruct.Message.processLeavingTail(Message.java:264)
  at net.sf.saxon.expr.instruct.Block.processLeavingTail(Block.java:660)
  at net.sf.saxon.expr.instruct.Template.applyLeavingTail(Template.java:239)
  at net.sf.saxon.trans.Mode.applyTemplates(Mode.java:1057)
  at net.sf.saxon.Controller.transformDocument(Controller.java:2088)
  at net.sf.saxon.Controller.transform(Controller.java:1911)
  ...

The problem is reported at: Exception during execution in Saxon-HE-9.5.1-6 and also tracked at https://saxonica.plan.io/issues/2104.

Update: according to Michael Kay the issue is fixed. See note #4:

I have committed a patch to Chain.itemAt() on the 9.5 and 9.6 branches to check for index<0

Enumerable class contains many overloads with IEqualityComparable<T> argument. Most notable methods are:

• Contains;
• Distinct;
• Except;
• GroupBy;
• Intersect;
• Join;
• ToDictionary;
• ToLookup;
• Union.

Recently we dealt with simple case:

source.
  Select(
    item =>
      new Word
      {
        Text = ...,
        LangID = ...,
        Properties = ...
        ...
      }).
  Distinct(equality comparer by Text and LangID);

In other words how do you produce a enumeration of distinct words from a enumeration of words, where two words are qualified equal if their Text and LangID are equal?

It turns out it's cumbersome to implement IEqualityComparer<T> interface (and any other interface in C#), at least it's nothing close to a conciseness of lambda functions.

Here we've decided to step in into framework space and to introduce an API to define simple equality comparers for a class.

We start from the use case:

var wordComparer = KeyEqualityComparer.Null<Word>().
  ThenBy(item => item.Text).
  ThenBy(item => item.LangID);

...
source.Select(...).Distinct(wordComparer);

And then proceed to the API:

namespace NesterovskyBros.Linq
{
  using System;
  using System.Collections;
  using System.Collections.Generic;

  /// <summary>
  /// A equality comparer extensions.
  /// </summary>
  public static class KeyEqualityComparer
  {
    /// <summary>
    /// Gets null as equality comparer for a type.
    /// </summary>
    /// <typeparam name="T">A type.</typeparam>
    /// <returns>
    /// null as equality comparer for a type.
    /// </returns>
    public static IEqualityComparer<T> Null<T>()
    {
      return null;
    }

    /// <summary>
    /// Creates an equality comparer for a enumeration item.
    /// </summary>
    /// <typeparam name="T">A type.</typeparam>
    /// <param name="source">A source items.</param>
    /// <param name="keyFunc">A key function.</param>
    /// <returns>
    /// null as equality comparer for a type.
    /// </returns>
    public static IEqualityComparer<T> EqualityComparerBy<T, K>(
      this IEnumerable<T> source,
      Func<T, K> keyFunc)
    {
      return new KeyEqualityComparer<T, K>(keyFunc);
    }

    /// <summary>
    /// Creates an equality comparer that uses this comparer as a base.
    /// </summary>
    /// <typeparam name="T">A type.</typeparam>
    /// <typeparam name="K">A key type.</typeparam>
    /// <param name="equalityComparer">A base equality comparer.</param>
    /// <param name="keyFunc">A key function.</param>
    /// <returns>
    /// An equality comparer that uses this comparer as a base.
    /// </returns>
    public static KeyEqualityComparer<T, K> ThenBy<T, K>(
      this IEqualityComparer<T> equalityComparer,
      Func<T, K> keyFunc)
    {
      return new KeyEqualityComparer<T, K>(keyFunc, equalityComparer);
    }
  }

  /// <summary>
  /// Equality comparer that uses a function to extract a comparision key.
  /// </summary>
  /// <typeparam name="T">A type.</typeparam>
  /// <typeparam name="K">A key type.</typeparam>
  public struct KeyEqualityComparer<T, K>: IEqualityComparer<T>
  {
    /// <summary>
    /// Creates an equality comparer.
    /// </summary>
    /// <param name="keyFunc">A key function.</param>
    /// <param name="equalityComparer">A base equality comparer.</param>
    public KeyEqualityComparer(
      Func<T, K> keyFunc,
      IEqualityComparer<T> equalityComparer = null)
    {
      KeyFunc = keyFunc;
      EqualityComparer = equalityComparer;
    }

    /// </summary>
    /// <param name="x">The first object of type T to compare.</param>
    /// <param name="y">The second object of type T to compare.</param>
    /// <returns>
    /// true if the specified objects are equal; otherwise, false.
    /// </returns>
    public bool Equals(T x, T y)
    {
      return ((EqualityComparer == null) || EqualityComparer.Equals(x, y)) &&
        EqualityComparer<K>.Default.Equals(KeyFunc(x), KeyFunc(y));
    }

    /// <summary>
    /// Returns a hash code for the specified object.
    /// </summary>
    /// <param name="obj">
    /// The value for which a hash code is to be returned.
    /// </param>
    /// <returns>A hash code for the specified object.</returns>
    public int GetHashCode(T obj)
    {
      var hash = EqualityComparer<K>.Default.GetHashCode(KeyFunc(obj));

      if (EqualityComparer != null)
      {
        var hash2 = EqualityComparer.GetHashCode(obj);

        hash ^= (hash2 << 5) + hash2;
      }

      return hash;
    }

    /// <summary>
    /// A key function.
    /// </summary>
    public readonly Func<T, K> KeyFunc;

    /// <summary>
    /// Optional base equality comparer.
    /// </summary>
    public readonly IEqualityComparer<T> EqualityComparer;
  }
}

So, now you can easily build simple equality comparers to cache them or instantiate on the fly. This comparers are usually related to property values or their function of source values.

See also LINQ extensions

Dasha Nesterovsky

This is a small post about refactoring lock statements in async methods.

Before refactoring we had a code like this:

lock(sync)
{
  result = methodToRefactorIntoAsync();
}

...

private object sync = new object();

Lock is bound to a thread, thus no way you to use it in async code. As an alternative you may use SemaphoreSlim class:

await sync.WaitAsync(cancellationToken);

try
{
  result = await methodAsync(cancellationToken);
}
finally
{
  sync.Release();
}

...

private SemaphoreSlim sync = new SemaphoreSlim(1, 1);


What will you do if you have async Web API method that runs on server for a some time but your client is dropped?

There are two solutions:

1. Run method to the end and allow to a framework to deal with disconnect;
2. Try to be notified about client's drop and to break early.

The first approach is simplest but might result in some overconsumption of server resources. The other method requires you to check client status from time to time.

Fortunatelly, ASP.NET provides a HttpResponse.ClientDisconnectedToken property, which is limited to IIS 7.5+ in integrated mode, but still fits our needs. So, you should request ClientDisconnectedToken, if any, and implement your async code using that token.

The following extension function gets that token:

using System.Linq;
using System.Net.Http;
using System.Threading.Tasks;
using System.Threading;
using System.Web;

public static class HttpApiExtensions
{
  public static CancellationToken GetCancellationToken(
    this HttpRequestMessage request)
  {
    CancellationToken cancellationToken;
    object value;
    var key = typeof(HttpApiExtensions).Namespace + ":CancellationToken";

    if (request.Properties.TryGetValue(key, out value))
    {
      return (CancellationToken)value;
    }

    var httpContext = HttpContext.Current;

    if (httpContext != null)
    {
      var httpResponse = httpContext.Response;

      if (httpResponse != null)
      {
        try
        {
          cancellationToken = httpResponse.ClientDisconnectedToken;
        }
        catch
        {
          // Do not support cancellation.
        }
      }
    }

    request.Properties[key] = cancellationToken;

    return cancellationToken;
  }
}

And here is a Web API WordCount service described in the previous post:

using System;
using System.Collections.Generic;
using System.Linq;
using System.Net;
using System.Net.Http;
using System.Threading;
using System.Threading.Tasks;

public class ValuesController: ApiController
{
  public async Task<int> GetWordCount([FromUri(Name = "url")] string[] urls)
  {
    var cancellationToken = Request.GetCancellationToken();

    using(var client = new HttpClient())
    {
      return (await Task.WhenAll(
        urls.Select(url => WordCountAsync(client, url, cancellationToken)))).Sum();
    }
  }

  public static async Task<int> WordCountAsync(
    HttpClient client,
    string url,
    CancellationToken cancellationToken)
  {
    string content = await (await client.GetAsync(url, cancellationToken)).
      Content.ReadAsStringAsync();

    return WordCount(content);
  }

  private static int WordCount(string text)
  {
    var count = 0;
    var space = true;

    for (var i = 0; i < text.Length; ++i)
    {
      if (space != char.IsWhiteSpace(text[i]))
      {
        space = !space;

        if (!space)
        {
          ++count;
        }
      }
    }

    return count;
  }
}

Though is simple there is a nuisance. You should pass cancellation token here and there, which adds to a pollution from async.

Though parallel and async algorithms solve different tasks, they converge in some cases. And it's not always immediately clear what's the best.

Consider the following task: get a total word count contained in a given a set of urls.

At first we've solved it as a parallel task: indeed this fits to MapReduce pattern when you get urls' contents to count the number of words in  parallel (Map), and then sum word counts per each url to get final result (Reduce). But then we decided that the very same MapReduce algorithm can be implemented with async.

This is a parallel word count:

public static int ParallelWordCount(IEnumerable<string> urls)
{
  var result = 0;

  Parallel.ForEach(
    urls,
    url =>
    {
      string content;

      using(var client = new WebClient())
      {
        content = client.DownloadString(url);
      }

      var count = WordCount(content);

      Interlocked.Add(ref result, count);
    });

  return result;
}

Here is async word count:

public static async Task<int> WordCountAsync(IEnumerable<string> urls)
{
  return (await Task.WhenAll(urls.Select(url => WordCountAsync(url)))).Sum();
}

public static async Task<int> WordCountAsync(string url)
{
  string content;

  using(var client = new WebClient())
  {
    content = await client.DownloadStringTaskAsync(url);
  }

  return WordCount(content);
}

And this is an implementation of word count for a text (it's less important for this discussion):

public static int WordCount(string text)
{
  var count = 0;
  var space = true;

  for(var i = 0; i < text.Length; ++i)
  {
    if (space != char.IsWhiteSpace(text[i]))
    {
      space = !space;

      if (!space)
      {
        ++count;
      }
    }
  }

  return count;
}

Our impressions are:

1. The parallel version is contained in one method, while the async one is implemeneted with two methods.

   This is due to the fact that C# compiler fails to generate async labmda function. We attribute this to Microsoft who leads and implements C# spec. Features should be composable. If one can implement a method as a lambda function, and one can implement a method as async then one should be able to implement a method as an async lambda function.

2. Both parallel and async versions are using thread pool to run their logic.

3. While both implementations follow MapReduce pattern, we can see that async version is much more scaleable. It's because of parallel threads stay blocked while waiting for an http response. On the other hand async tasks are not bound to any thread and are just not running while waiting for I/O.

This sample helped us to answer the question as to when to use parallel and when async. The simple answer goes like this:

• if your logic is only CPU bound then use parallel API;
• otherwise use async API (this accounts I/O waits).

Not a long ago C# has introduced special language constructs to simplify asynchronous programming. It seems C++1x will follow async trend. But only recently when frameworks like ASP.NET Web API and Entity Framework started to catch up we've felt how it's to program with async and await keywords.

At first glance it seems it's a pure pleasure to write async methods:

private async Task SumPageSizesAsync()
{
  // To use the HttpClient type in desktop apps, you must include a using directive and add a
  // reference for the System.Net.Http namespace.
  HttpClient client = new HttpClient();
  // . . .

  byte[] urlContents = await client.GetByteArrayAsync(url);
  // . . .
}

To dereference a Task<T> into T you just write await task_t_expression, mark your method with async specifier, and adjust output type (if not void) to Task or Task<Result>. Compiler applies its magic to convert your code into an asynchronous state machine.

We liked this feature and immediately have started to use it. But, as we said, async/await has shined in full when frameworks made it a core element, and at that point we have started to see that while  async/await solve the task, they does not abstract the developer from implementation details, as a code gets considerably polluted.

Consider a method with pollution marked:

public static async Task<UserAuthorization> GetAuthorizationAsync(string accessToken)
{
  var key = "oauth2:" + accessToken;
  var authorization = cache.Get<UserAuthorization>(key);

  if (authorization != null)
  {
    return authorization;
  }

  using(var model = new ModelContainer())
  {
    authorization =
      (await model.UserAuthorizations.
        Where(item => item.AccessToken == accessToken).
        ToListAsync()).
      FirstOrDefault();
  }

  if (authorization == null)
  {
    authorization = await ValidateAsync(accessToken);
  }

  cache.Set(key, cache.ShortDelay, authorization);

  return authorization;
}

The more you use async, the more pollution you will see in your code. Ideally we would like to see the method quoted without any marked parts.